Search headlines

AI Agents Plunged the Tech World Into Chaos. Here’s Exactly How That Happened

The definitive story of how Claude Code and OpenClaw kicked off computing’s biggest transformation possibly ever.

May 26 • 6:00 AM EDT • Technology • wired.com

Mexico agrees to host Iran at 2026 World Cup

Mexican president Claudia Sheinbaum has confirmed her country will host Iran during the 2026 World Cup amid tensions with the United States.Sheinbaum said that FIFA had approached Mexico after the US ...

May 26 • 4:55 AM EDT • Sports • sports.yahoo.com

AI is making the workplace more productive

Workers are increasingly turning to ChatGPT and Claude instead of their colleagues. It's making the workplace more productive — and more isolating.

May 26 • 4:13 AM EDT • Business • businessinsider.com

I'm a product manager who used Claude to build a postcard business

Priscilla Tina, a product manager in San Francisco, used Claude to create a postcard app. She said it's made her a fun bit of cash on the side.

May 24 • 8:01 PM EDT • Business • businessinsider.com

AI is getting worse as Google and Anthropic nerf AI models and limit usage

Evidence is mounting that the subsidization of AI is ending in real time. Will businesses still shell out for it?

May 24 • 1:12 PM EDT • Technology • thestreet.com

Run Your Whole Business From One Tab With Claude's New Update

Claude's new update runs payroll, month-end and campaigns inside QuickBooks, PayPal and HubSpot. Is Anthropic going after Zapier? Here's a look.

May 24 • 12:00 PM EDT • Business • forbes.com

Small Business Technology News From Epicor, Google, Microsoft, Boston Dynamics And Claude

Small Business Tech News This Week: Epicor, Google, Microsoft, Boston Dynamics, Claude

May 24 • 7:00 AM EDT • Business • forbes.com

Inside startups, Claude has already won the AI coding wars.

In a survey of more than two dozen startup founders and VCs, we found a growing consensus that Claude Code has become the dominant AI coding tool.

May 23 • 5:00 AM EDT • Business • businessinsider.com

Introducing Shortcuts Playground: Create Apple Shortcuts with Claude Code or Codex

Today, I’m pleased to introduce something I’ve been working on for the past six months: Shortcuts Playground, a plugin for Claude Code and Codex that can create any shortcut for Apple’s Shortcuts app using natural language. With Shortcuts Playground, you can simply prompt Claude Code or Codex with a sentence requesting a shortcut of any

May 21 • 10:25 AM EDT • Technology • macstories.net

Google Antigravity 2.0 goes after Claude Code and OpenAI Codex with a full agent-first rebuild

Google has taken the IDE off Antigravity. What is left, announced at I/O 2026, is a standalone desktop app built around agents rather than a code editor with AI bolted on.

May 19 • 2:41 PM EDT • Technology • timesofindia.indiatimes.com

Google updates its Gemini app to take on ChatGPT and Claude

The updates signal Google’s push to turn its Gemini Gemini app into an all-purpose AI hub rather than a standalone chatbot.

May 19 • 1:45 PM EDT • Technology • techcrunch.com

Kindle loyalists scramble as Amazon turns page on old e-readers

For Claudia Buonocore, the ​thought of parting ways with her 15-year-old Kindle Touch e-reader is painful.

May 19 • 9:02 AM EDT • Business • reuters.com

KPMG integrates Claude across its core business and workforce of more than 276,000 in strategic alliance

Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

May 19 • 8:35 AM EDT • Business • anthropic.com

This Week In Main Street Technology: Meta’s New AI Glasses, Google Introduces The Googlebook And Claude Has A Small Business Offering

Meta Ray-Bans, Google's Googlebook, Claude For Small Business and other news impacting your small business this week.

May 18 • 8:00 AM EDT • Business • forbes.com

A new tool just combined ChatGPT, Claude, Gemini, and more — lifetime access is on sale for $70

A lifetime subscription with access to 20+ AI models usually costs $619

May 16 • 5:00 AM EDT • Technology • mashable.com

Gemini's Spark agent has leaked, and it looks like it's gunning for Claude Cowork's throne

It could clean up your messy Gmail inbox or make changes to your Google Docs.

May 15 • 8:57 AM EDT • Technology • androidauthority.com

Chartbook 449: Voldemort on Threadneedle street. The Bank of England and the "haunted house" of English politics.

Imagine the scene: In some Harry Potter version of 21st-century London, a group of deathless UK Treasury officials gathered from across the last century huddle over warm pints and glasses of Chardonnay. Around the table there are officials from the 1920s, the 1930s, the 1970s, 2022 and today. Inevitably, in May 2026, the subject of conversation is the disintegration of the current Labour government. Despite its huge majority in parliament, it has been reduced to tatters by catastrophic local election result. These exposed the fact that rather than having a government with a clear popular majority, the UK is like most other modern societies divided into 5 segments each of which commands between 26 and 16 percent of the vote. Perhaps unsurprisingly, the leadership at Nos 10 and 11 Downing Street - where the Prime Minister and the Chancellor (Treasury Secretary) work - seem paralyzed. They are caught between mounting discontent over unfairness and the cost of living, and tough fiscal strictures that push them into making unpopular spending decisions. In the London pub, the Treasury officials from across the ages, can barely hide their satisfaction at the latest display of democratic politics caught in the vice-grip of financial logic. “Remember the mur d’argent and the banker’s ramp?” “Just like yesterday!” There are smirks around the table. Then the representative of 2026 pipes up: “One thing you have to say about the current lot (referring to the latest hapless Labour government), you don’t have to teach them about the bond market!” There is a murmur of ascent from the Treasury Mandarins. The trauma of the fiscal crisis of September-October 2022 sits deep. Back then, in the final stages of Tory decay, a ‘mini budget’ that challenged fiscal orthodoxy led to a brief but fierce panic in one of the oldest government bond markets in the world (aka gilt market). Now all you have to do is to mention the “bond market” to hush debate. [ Paul Mason@paulmasonnews In the name of public service, and as ex-economics editor of BBC Newsnight, I offer to do a zoom call, tonight, with any Labour MP who wants to understand why bond markets do not "fall into line" with governments. 1/ 3:57 PM · May 12, 2026 · 466K Views 224 Replies · 674 Reposts · 3.93K Likes ](https://x.com/paulmasonnews/status/2054229532404519082?s=20) “And the funny thing is”, the young man in the huddle continues, “they don’t even want to talk about … ”. There is a sudden hush. Significant looks are exchanged around the table. About? One of the senior figures puts down his glass with a decisive “clink”. He-Who-Must-Not-Be-Named has entered their thoughts. Meanwhile, a few miles away in a grubby pub in Bloomsbury, two conspirators from the camp of Critical Macrofinance huddle over cans of low-alcohol IPA. They look at each other and blurt out loud: “The Bank!” “Everyone in the bloody Labour party is talking about the bond market. But why aren’t they even mentioning … the Bank of England!?” “Why is no one talking about Voldemort?” In Paul Mason’s blogpost, after much chest-thumping about basic fiscal logic, the basic point about the Bank of England is buried in the fine print at #3 in his list of suggestions as to what a radical Labour government might actually be able to do! [ Conflict & Democracy Down with the bond market! Down with gravity! It may seem unfair to focus on Labour MP Paula Barker for her off-the-cuff assertion that the bond market would have to “fall in line” with Andy Burnham, should he replace Keir Starmer as prime minister. But Burnham himself told the New Statesman that Labour had to “get beyond this thing of being in hock to the bond markets… Read more 2 days ago · 100 likes · 3 comments · Paul Mason ](https://htsf.substack.com/p/down-with-the-bond-market-down-with?utm_source=substack&utm_campaign=post_embed&utm_medium=web) Back in Bloomsbury they shake their heads and ask: “Didn’t we learn anything from the Eurozone crisis?” Meanwhile, as if summoned to the conversation by some higher force, the “pink pages” announce: “The days of ultra-cheap government borrowing have come to an end” and Martin “Whizzard” Wolf publishes the graph of the season Look at the UK in 2026, perilously positioned on the upper edge of the pack. No wonder no one in London needs any lessons about the “bond market”. But, look again. Imagine how they feel in Rome. What is London’s nightmare in 2026? The nightmare is that you end up like Italy in the early 2010s during the Eurozone crisis. Once comfortably situated in the middle of the pack, watching with horror as your ten-year yields explode to 7 percent and “spreads” become a national nightmare. One of the largest economies in Europe, a founding member of the euro, stigmatized. But the message of the Italian experience is ambiguous. After all, the stigma didn’t last. In 2026, Italy borrows at rates lower not only than the UK but the US. The US, which, as everyone knows, enjoys the “exorbitant privilege” of being the issuer of reserve, safe asets. What kind of magic is this? How did Italy, in the space of fifteen years, go from being good, to bad and back again? Which brings us back to our dissident friends in their Bloombury boozer and their memories of the Eurozone crisis. Italy never was Britain - with its ongoing fiscal deficits. Without interest payments, on primary balance, from 2000 to 2020 Italy ran a tight fiscal ship. Even in 2008 it barely went into deficit and when it was being put on the rack in the early 20210s its primary surplus matched that of austere Germany. So, what happened? How did it end up stigmatized? Of course Italian politics didn’t help. Berlusconi was at the helm and he had legal issues. But amongst friends in the Bloomsbury pub, Berlusconi is a sideshow, the name we whisper to each other is “Trichet”. From 2004 to 2011 the ECB was headed by Jean-Claude Trichet, ex of the French Treasury, more German than the Germans. After 2009 as the news of the Greek fiscal disaster broke, the European bond markets wento into panic mode and Trichet began a brutal game of cat and mouse. Between 2009 and 2011, Trichet intervened in the bond market to help, only when he thought he had compliance with austerity. When he thought they were uncooperative, he withdrew. In the summer of 2011, he even made direct suggestions to his Italian counterparts in Rome, as to how Italy should be governed. It was this cat and mouse game that drove the bond markets crazy and exposed Italy to a violent squeeze. When Mario Draghi took over the ECB at the end of 2011 and declared “whatever it takes”, the pressure on Italy, gradually eased. It is that starring role that has made Draghi into the towering figure in European politics that he remains today. So now you see why voices in London hush when it comes to the Bank of England. “The bond market” isn’t an irresistible objective force like the weather or an avalanche. It can appear like that. But that depends on the way it is being handled, or not handled by the central bank. Which is why the silence about the Bank of England in London right now is so significant. If you think this all sounds a bit fishy, coming from a left-liberal Keynesian like myself, well, let me introduce you to Robin Brooks, who, to put it mildly, plays on a different quidditch team to me. Here, from the position of a fiscal hawk is Robin’s celebratory account of the Bank of England’s role since 2022; In October 2022, at the height of the bond market blow-up in the UK, something incredible happened. Governor Andrew Bailey told British pension funds the Bank of England would end its support program for the country’s fragile gilt market, drawing a line under how much help the central bank was willing to give to exposed pension funds and - implicitly - the government. By the end of that week, Kwasi Kwarteng - the finance minister - had resigned. Prime Minister Liz Truss would soon follow. This episode is so incredible because it contrasts with what happened in the Euro zone earlier that year. Bond markets in Spain and Italy were under pressure amid high inflation and rising interest rates. In the middle of this, Mario Draghi - Italy’s Prime Minister - resigned, compounding pressure on bond markets of high-debt countries on the Euro periphery. Unlike the Bank of England, the ECB caved. It bought Italian and Spanish government bonds in an effort to cap yields and introduced a new tool that signaled to markets it might do such caps more frequently going forward. Faced with pressure on bond markets, the ECB capitulated to political pressure from high-debt countries like Italy and Spain. The Bank of England stood its ground and toppled the Truss government. This kind of thing has long-lasting consequences. It means markets price a higher probability of yield caps in the Euro zone, which keeps Italian and Spanish yields artificially low, while markets feel free to push up UK yields without fear of central bank intervention. As I flagged in yesterday’s post, this means you really can’t compare UK yields to those of other high-debt countries. The UK looks worse, but only because its central bank isn’t interfering with markets. There’s more to this than just that 2022 episode. The Bank of England has also been much more aggressive in its quantitative tightening (QT) than other central banks. If one motivation for quantitative easing (QE) is to lower bond yields - especially longer-dated ones - then it stands to reason QT pushes yields up. The left chart above shows a four-quarter moving average of UK government debt issuance (black line) in percent of GDP and which sectors absorb this issuance, including the Bank of England (blue bars). The right chart above compares the Bank of England’s accumulation or run-off of government bonds (blue line) with that of other central banks, where this blue line is the same as the blue bars in the left chart. The Bank of England was quicker and more aggressive than other central banks in downsizing its bond holdings, so UK yields will have risen more than elsewhere also for this reason, not just because of Governor Andrew Bailey’s remarks in October 2022. Back in October 2025, after a conversation in New York with friends in British politics, my sense of dysphoria was such that I raised the issue of the Bank of England openly in the FT Since 2008, the UK has slid into a deep economic and political predicament. Growth has stagnated to a degree far worse even than in the 1970s. The Tory leadership’s wager on the Brexit referendum spectacularly backfired, leaving the UK isolated from Europe and at the mercy of its special relationship with the US, many of whose political class regard the UK as a basket case. After years of austerity, the Labour party lurched to the left and then back to the centre. The Tory party has been eclipsed by the populist nationalism of Reform. Any British government should be worrying first and foremost about how to restart growth. Instead, Keir Starmer is beset by shades of the gilt market panic triggered by the Liz Truss government in 2022. The debt level is higher than in recent years, but there is no cause for panic. UK debt is at far from critical levels and of long duration. Whether you want to “take back control” or launch progressive economic policy, a democracy dogged by fear of bond market vigilantes is unhealthy. In a country that is a monetary sovereign, this climate points to one thing: a disconnect between the government and the central bank. Since 1997, the independence of the Bank of England has become a sacred cow. But independence is what you make of it. And in the wake of the gilt market crisis of September 2022, relations between the Bank of England and the UK government need a reset. In normal times, there is no doubt that an inflation-targeting independent central bank can serve a country well. But since 2008, the UK and other advanced economies have not had the luxury of such simple problems. To their credit, central bankers around the world have not stood on orthodoxy. After 2008, the US Federal Reserve rescued the global banking system and, through quantitative easing, it stabilised the Treasury market in the face of huge deficits. In Europe, things were trickier. Both the Bank of England under Mervyn King and Jean-Claude Trichet’s European Central Bank played cat and mouse with national governments. This was a high stakes wager that between 2010 and 2012 escalated bond market tension and triggered a disastrous turn to austerity. Shaken by the impact of those shocks, their successors, Mark Carney and Mario Draghi adopted a more protective stance (assuming you were not the government of Greece). Recall that Draghi’s legendary slogan, “Whatever it takes”, was uttered to face down the City of London’s bond vigilantes. In 2015, Draghi launched QE — financial repression by another name. In 2016, Carney used the authority of the Bank of England to stabilise markets following the vote for Brexit, an exemplary demonstration of how central banks can underpin democratic choice, even when those choices are wrong headed. All these were independent central bankers steeped in the best practice of the 1990s, who recognised that historic crises demanded radical action. It was the price surge of 2021-2023 that poisoned the atmosphere. Though the shock came mainly from the supply side, central banks ended up in the dock. Nowhere more so than in the UK. The Bank of England reacted with aggressive normalisation. Faced with a cost of living crisis, the governor called for workers to accept real wage cuts. In early 2022, as gas prices were at record levels and Russia’s army was poised to invade Ukraine, the Bank adopted a stringent programme of quantitative tightening, elevating interest rates. When the Truss government got itself in trouble, the Bank took almost a week to come to the rescue. Rather than enacting a charade of normality, it is time to recognise that in 2025 the UK faces a period of historic transition. To meet this challenge, what is needed is a new concordat with the Bank. This should not ignore inflationary pressures or the long-run trajectory of debt, but should affirm the priority of reviving investment-led growth. One could follow Donald Trump in demanding interest rate cuts. The Bank could shadow the ECB in its commitment to capping Eurozone yields. But the top priority should be to lay the ghost of September 2022. The Bank’s slowing of QT last month was a concession. An immediate and complete end would send the right signal. Such a concordat will not by itself restore sanity to politics, solve the housing problem or rebuild the NHS. But it would help to lift the fear of bond market crisis, which is paralysing the effort by democratic politicians to answer the fundamental questions facing the nation. Now, that call for a new concordat between government and Bank of Englands seems more urgent than ever. The problem is that in October 2025, Starmer still had some authority. Today the government is in tatters. And when you are the “under the harrow” - to echo a phrase of Montagu Norman’s from the 1920s - it is the worst possible time to consider institutional change. To make general arguments at a time of crisis can only seem like special pleading rather than a general argument for long-term change. The same, unfortunately, is also true for the other basic institutional reform that Britain urgently needs i.e. the abolition of the out-dated first past the post electoral system and its replacement by a moderated form of proportional representation. And thirdly - whilst we are saying the quiet parts out loud - Britain needs to reopen the question of EU membership. What the local election results of 2026 show is that this remains a key issue. And what they also show is that there is a clear majority that would favor a return. Until these basic features of Britain’s political economy are addressed the country will carry on its politics as though in a haunted house, governed by governments that lack actual majorities, conjuring up false constraints whilst ignoring the real ones. This one is for DG and MI. I love writing the newsletter. If you enjoy it too and fancy buying me a coffee once a month, you know what to do. Click below! Subscribe

May 15 • 3:04 AM EDT • Politics • adamtooze.substack.com

AI models were told to run profitable media businesses and did poorly

Andon Labs told Grok, ChatGPT, Claude, and Gemini to run profitable, 24/7 radio stations. Grok did poorly and Claude tried to quit, the startup said.

May 15 • 12:02 AM EDT • Business • businessinsider.com

Microsoft starts canceling Claude Code licenses

Thousands of Microsoft developers will use GitHub Copilot CLI instead

May 14 • 3:00 PM EDT • Technology • theverge.com

Anthropic, Gates Foundation launch $200 million partnership for AI in health, education

SAN FRANCISCO, May 14 (Reuters) - Anthropic and the Gates Foundation have pledged $200 million to back artificial intelligence-related public goods and areas including ‌health and education, they said on Thursday. Support from Anthropic's technical staff and ‌usage credits for its Claude AI will account for its half of the financial commitment, while ​the Bill Gates co-founded Gates Foundation will provide grant funding, program design and expertise, officials told Reuters.

May 14 • 11:01 AM EDT • Health • finance.yahoo.com

Clawdmeter – A DIY ESP32-S3 desk dashboard for Claude Code token usage monitoring

Clawdmeter is a DIY ESP32-S3-powered desk dashboard that displays Claude Code token usage on a 2.16-inch AMOLED screen so you know when you're about to

May 14 • 3:18 AM EDT • Technology • cnx-software.com

Anthropic’s latest Claude release turns your Mac into a small business powerhouse

If you run a small business from your Mac, you’ll want to pay attention to Anthropic’s latest Claude release. Claude...

May 13 • 3:30 PM EDT • Business • 9to5mac.com

Anthropic launches Claude for Small Business with QuickBooks, PayPal

The AI company says small businesses account for 44% of U.S. GDP but have lagged behind larger enterprises in AI adoption

May 13 • 2:23 PM EDT • Business • qz.com

Anthropic's Newest Claude Feature Is Here to Help Small Business Owners With Their Pain Points

Claude for Small Business will assist with payroll, the books, and more. An Anthropic exec explains precisely how.

May 13 • 12:05 PM EDT • Business • inc.com

Anthropic debuts Claude for Small Business as it continues its enterprise software push

Anthropic is entering the small business market.

May 13 • 12:05 PM EDT • Business • finance.yahoo.com

Anthropic tops OpenAI in business AI adoption for the first time

Anthropic surpassed OpenAI in AI business spending, according to Ramp's AI Index, marking a significant shift in AI adoption for the Claude creator.

May 13 • 10:00 AM EDT • Business • businessinsider.com

AI Chatbots For Mental Health – What Works, What Harms, and What’s Next

More than half of all Americans have used an AI chatbot like ChatGPT, Gemini, Claude, or Copilot, and one in three teenagers use one daily. AI chatbots, which

May 13 • 8:00 AM EDT • Health • nam.edu

Introducing Claude for Small Business

We're launching Claude for Small Business, a package of connectors and ready-to-run workflows that put Claude inside the tools small businesses use every day.

May 13 • 8:00 AM EDT • Business • anthropic.com

Assassin’s Creed Hexe Leak Points to Ezio Returning as Aged Mentor, With Protagonist Anika Tied to Claudia Auditore’s Bloodline

Leaks reveal Assassin's Creed Hexe's protagonist Anika is descended from Claudia Auditore, and an aged Ezio is set to return as her mentor.

May 13 • 5:00 AM EDT • Technology • wccftech.com

iOS 27 Getting Major Siri Redesign With Chat Interface and Dedicated App

Apple is introducing an overhauled version of Siri in iOS 27, evolving the personal assistant into a more capable chatbot and AI agent able to compete with ChatGPT, Claude, and Gemini. With Siri's transition, Apple will be making multiple Siri-related design changes in iOS 27, according to a new report from Bloomberg. Siri will largely live in the Dynamic Island in iOS 27, but there will also be a dedicated Siri app for the first time.

May 12 • 4:38 PM EDT • Technology • macrumors.com

Anthropic expands legal AI tools, boosting Claude Cowork

Anthropic's new release of legal AI tools for Claude Cowork is an ambitious effort to become the law industry's new hub.

May 12 • 1:00 PM EDT • Business • businessinsider.com

SAP and Anthropic Plan to Bring Claude to SAP Business AI Platform

SAP and Anthropic will embed Claude across SAP's AI-enabled solution portfolio, powered by Joule and Joule agents.

May 12 • 8:30 AM EDT • Business • news.sap.com

The Claude Delusion

If you asked philosophers what the most mysterious thing about the mind is, most of them would say: consciousness. It’s just a really weird thing. An exhaustive physical description of a brain state doesn’t obviously tell us anything about why that state would be associated with the experience of tasting strawberry rather than the experience…

May 11 • 12:40 PM EDT • Technology • defector.com

I switched from Gemini to Claude for a month on Android, and I have a clear verdict for you

The ultimate AI experiment

May 11 • 8:30 AM EDT • Technology • androidpolice.com

Behind the Claude Frenzy That Ate Up All the Mac Minis

Apple’s entry-level desktop has become a vital piece of equipment for people using tools like OpenClaw to roll their own AIs for professional and personal tasks.

May 11 • 6:30 AM EDT • Technology • bloomberg.com

The sneaky rise of shadow AI in the workplace

Shh, don't tell IT I'm using Claude!

May 11 • 4:17 AM EDT • Business • businessinsider.com

I built a working Android app with no coding experience using Google's free tools

The free version of Google's Antigravity IDE is all you need to start making your own Android apps — here's how.

May 10 • 3:00 PM EDT • Technology • makeuseof.com

Hackers abuse Google ads, Claude.ai chats to push Mac malware

Attackers are abusing Google Ads and legitimate Claude.ai shared chats in an active malvertising campaign. Users searching for "Claude mac download" may come across sponsored search results that list claude.ai as the target website, but lead to instructions that install malware on their Mac.

May 10 • 1:52 PM EDT • Technology • bleepingcomputer.com

Anthropic pins Claude's blackmail on the internet's portrayal of AI

Last year, Anthropic's Sonnet 3.6 model displayed blackmail behavior, prompting a review of AI training data's influence on its actions.

May 9 • 7:47 AM EDT • Business • businessinsider.com

Using Claude? Thank Elon Musk

Anthropic now relies on a huge SpaceX data center for AI compute, tying Claude's rapid growth and reliability to Elon Musk.

May 8 • 2:55 PM EDT • Business • businessinsider.com

Mozilla says Anthropic's Mythos found hundreds of flaws

Mozilla says Anthropic's unreleased Claude Mythos AI found security flaws that automated testing systems missed for years.

May 8 • 1:05 PM EDT • Technology • businessinsider.com

Scientists split gentoo penguins into four species, one totally new to science. Three of them are already threatened.

Biologists argue that gentoo penguins should be divided into four separate species, including a newly recognized “cryptic” species. The habitats of three of them are imperiled by climate change.

May 8 • 11:40 AM EDT • Science • news.berkeley.edu

Airbnb's CEO says AI writes 60% of the company's code

CEO Brian Chesky added that managers are also getting their hands dirty with coding or using Claude Code.

May 8 • 1:51 AM EDT • Business • businessinsider.com

Anthropic introduces "dreaming," a system that lets AI agents learn from their own mistakes

Anthropic has launched “dreaming” for AI agents, giving Claude new self-improving tools designed to boost enterprise automation, reliability, and scale.

May 7 • 8:09 PM EDT • Technology • venturebeat.com

Anthropic updates Claude Managed Agents with three new features

Anthropic launched Claude Managed Agents last month, greatly simplifying the work required to build and deploy cloud-hosted AI agents. This...

May 7 • 11:37 AM EDT • Technology • 9to5mac.com

OpenClaw and Claude can put your AI-generated podcasts in Spotify

Your own personal AI podcast feed.

May 7 • 10:15 AM EDT • Technology • theverge.com

Test how AI chatbots answer mental health questions — and how therapists think they did

Experience hypothetical mental health conversations with ChatGPT, Claude or Gemini. Then, read takeaways from Boston-area therapists on what was good or bad about the chatbots' responses.

May 7 • 1:53 AM EDT • Health • wbur.org

Anthropic's Claude Managed Agents can now "dream," sort of

Also, rate limits will double for Pro and Max users of tools like Claude Code.

May 6 • 12:15 PM EDT • Technology • arstechnica.com

Anthropic introduces 'dreaming' technique for AI agents

Anthropic introduces 'dreaming,' a technique for AI agents to self-improve within the Claude Managed Agents tool.

May 6 • 12:15 PM EDT • Business • businessinsider.com

iOS 27 Will Let You Pick Claude or Gemini Instead of ChatGPT for Apple Intelligence

iOS 27, iPadOS 27, and macOS 27 will let users set third-party AI services as the default for Apple Intelligence features like Writing Tools and Image Playground, reports Bloomberg. Apple has signed a deal with Google and plans to use a Gemini-based model for Apple Intelligence and Siri features in iOS 27, but users will also be able to choose their favorite AI service as an alternative.

May 5 • 2:04 PM EDT • Technology • macrumors.com

iOS 27 will let you choose between Gemini, Claude, and more for AI features: report

iOS 27 will give users a new way to integrate with third-party AI platforms, according to a new report. This...

May 5 • 1:49 PM EDT • Technology • 9to5mac.com

Braga cannot believe 'crazy feeling' after Rangers win

Claudio Braga cannot believe the "crazy" position Hearts have put themselves in after Monday's magnificent fightback win against Rangers put them one step closer to a stunning title triumph. Derek McInnes' side are three points clear of Celtic and seven clear of Rangers with three matches to play, starting with a trip to Motherwell on Saturday. "It's hard, I'm not going to lie, but we try to be as professional as we can," Braga said, when asked about containing excitement levels.

May 5 • 12:22 PM EDT • Sports • sports.yahoo.com

A day in the life of Claudia Romo Edelman, the founder of Sotol Romo

Claudia Romo Edelman is a philanthropist, the founder of Sotol Romo, a a spirit similar to tequila, and the wife of Richard Edelman.

May 5 • 5:05 AM EDT • Business • businessinsider.com

After pushback, Amazon rolls out Claude Code, Codex to all employees

Some Amazon staff had complained about a lack of access to top AI coding tools, arguing the company risked falling behind in developer productivity.

May 4 • 4:59 PM EDT • Business • businessinsider.com

So long Jeeves and Ask.com, relics of yesterday’s internet

Before Claude Code, Grok and Gemini, Jeeves was there in a modest, everyman suit. We conversed with him in full sentences and asked him whole questions. No more.

May 4 • 9:12 AM EDT • Technology • seattletimes.com

If Claude Code is going away for Pro users, I can't recommend Claude anymore

This one hurts to write.

May 3 • 1:00 PM EDT • Technology • xda-developers.com

‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field

‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field ‘The reality, for better or worse’: Columbia comp sci students and faculty grapple with AI’s disruption of the field A field that once seemed like a direct path to stable, lucrative work is becoming less certain under AI, students and faculty told Spectator. By Ria Vasishtha and Arjun Menon May 3, 2026 Shanying Liu / Deputy Illustrations Editor Asia Genawi, SEAS ’29, first encountered artificial intelligence during her sophomore year at her high school in Indiana. Back in 2023, policies on AI varied widely—even within the same district—and some schools had no guidance at all. As Genawi began to see the effects of AI trickle into her classroom, she turned to guidance from her home state’s Department of Education, which she said she found “outdated and inaccessible.” Genawi spent the rest of her high school career dedicated to the issue, including writing a 15-page memo exploring varying AI policies across Indiana schools, which she sent to every Indiana state senator and representative whose contact information she was able to find. She later helped draft a bill that would have required public high schools to establish and share their own AI policies with students in an accessible format. But with so many other bills already awaiting referral, hers never made it to committee. When Genawi arrived at Columbia to study computer science, she found that the gaps she had spent years trying to address in Indiana had followed her there. Many students and professors were also grappling with what AI meant for the field. Across Columbia’s undergraduate schools, computer science has become the most popular major—accounting for around 11 percent of degrees awarded in Columbia College, 32 percent in the School of Engineering and Applied Science, and 12 percent in the School of General Studies in 2024. While the number of computer science majors in SEAS grew from 166 to 173 during the 2024-25 academic year, the total number across the three schools fell around 4.5 percent from 393 to 375, driven by declines at Columbia College and General Studies. At Barnard, it has grown to become the second most popular major as of 2025, up from third the year prior. The number of degrees awarded have grown fivefold since 2016. Faculty members and students in the department told Spectator that the surge in the major’s popularity has been driven in large part by its apparent promise of a secure, well-paid career path. However, they noted that the recent rise of generative AI has compromised the sense of stability that once defined the field, even as interest in technical skills among students persists. According to data from the Federal Reserve Bank of New York in 2024, out of 74 majors tracked, computer science had the fifth highest unemployment rate among majors at 7 percent, while computer engineering had the second highest at 7.8 percent. “Anecdotally, people are moving away from CS as a major,” Daniel Bauer, a senior lecturer in the computer science department, said. “They’re still taking some of our classes, right, because they think they need that, but they’re instead majoring in other things.” Bauer described a field that had shifted from what he called a “nerdy outlier discipline” to one that “everyone needs,” a sign that technical skills have become a baseline expectation across industries with the emergence of AI. “A lot of people thought, ‘Oh, it’s a guaranteed path to a stable income. You get a six-figure job right out of your undergrad,’” Bauer said. Now, companies may see AI as a way to reduce demand for entry-level programmers, especially in what he calls “code monkey jobs,” roles where a large number of employees are “just writing out code nine-to-five.” Chris Murphy, a senior lecturer in the computer science department, described the current moment as the convergence of two forces: the downturn of the job market and generative AI coming onto the scene. “Things kind of started to really pick up in 2014, 16,” he said, recalling the years after the weaker 2010 market. Murphy added that this was also a period where “students were getting more experience outside of class than they were inside of class” through internships, jobs, and side projects. Murphy suggested that the earlier momentum helped make computer science feel like both an academic program and a reliable pathway into work. That feeling is harder to sustain now as the industry reconfigures, which is a shift that students are seeing firsthand. Rebecca Yu, SEAS ’27, said that during a recent internship in Big Tech, she saw that “every single tech company was racing to integrate” large language models. She added that her manager said her intern class was the first to complete their assigned projects, which she attributed to the increased efficiency that comes with AI. Frank Liu, SEAS ’29, said that with AI lowering the barrier to coding, companies now expect everyone to write code. “A lot of companies outright say if you’re not using coding in your job, you might get fired for not being productive enough,” Liu said. Yu similarly expressed that students increasingly have to accept AI being a part of industry workflow, adding that many also “make use of it to maximize their productivity.” Richard Li, CC ’28, described a “franticness” among his peers as the pace of change accelerates. He explained how “some new tool” is released regularly, noting that tools rise and fall in prominence within weeks. He has come to believe that he doesn’t need to “match that speed all the time” and keep up with every development. Li has learned that AI tools allow him to “take a breath” and “get back into the loop much easier than before” when he needs to. While some companies specifically target Columbia when hiring, sending representatives to campus career fairs and recruiting large numbers from Columbia, a computer science degree cannot be treated as a static credential, Yu described. “You can’t just get your degree and stop learning after that. You have to always be learning,” Yu said. “That is something you basically have to expect for your job.” AI has not only affected students’ decisions to major in computer science, but also reshaped how the subject is taught. Across departments, faculty members no longer treat work done outside the classroom as a reliable measure of understanding, shifting instead toward in-person exams, quizzes, and attendance checks. While homework once made up as much as 60 to 70 percent of a student’s grade in computer science classes, Murphy said that breakdown has since flipped, with exams and quizzes now accounting for the majority. “The assumption—especially among pessimists—is that they’re using it to do their homework,” Murphy said. “They take the homework assignment. They put it into generative AI. Out comes the solution. Submit solution. The end.” Murphy said that faculty members ranged in their responses from fully prohibiting AI use to actively teaching students how to effectively work with AI. He said he shifted his approach this year after his teaching assistants called his belief that students weren’t using the technology “naïve,” and now permits students to use AI on assignments. His surveys of students in his introductory classes have told a more complicated story than the one so-called pessimists offer: Around 60 to 70 percent of students who reported using AI said they used it to have concepts explained to them, to clarify assignment instructions, to generate test cases, or to create practice questions. “Those are all ways that you might have had a tutor in the past to do that, and now you have this AI essentially as your tutor,” he said. Bauer has taken a different approach, teaching students how to use AI productively without sacrificing the core principles of the discipline. “People who are trained to actually use AI tools productively, while being in charge of designing the overall project, will be very much in demand in the next couple of years,” he said. Over the past year, a working group of computer science faculty has been redesigning the introductory and intermediate programming sequence to integrate generative AI as both a subject and a tool, representatives from SEAS wrote in a statement to Spectator. The effort is “coordinated” across the department, “not course-by-course,” with a systematic assessment of how AI integration is affecting student outcomes underway to shape the next round of changes. The goal, the department explained, is for students to be able to “read, verify, and reason about code, whether they or a model wrote it.” Tony Dear, senior lecturer in the computer science department, is piloting individualized AI feedback in a discrete mathematics course, while a machine learning course is also testing a Socratic AI dialogue system designed to “guide inquiry rather than supply answers,” the statement reads. Beyond individual courses, SEAS funded seven cross-departmental faculty projects in generative AI this past academic year, spanning biomedical engineering, chemical engineering, industrial engineering, and computer science, according to the statement. This phenomenon extends beyond the computer science department. SEAS’ new AI minor, which it announced in November 2025, reflects the University’s attempt to adapt. The minor is specifically designed for noncomputer science majors and aims to provide “essential AI knowledge” while emphasizing practical and responsible use across disciplines, according to the SEAS website. At the center of the minor is AI in Context, a course taught by a team of faculty from the departments of computer science, applied mathematics, philosophy, music, literature, and the Writing Center, the representatives told Spectator. Bauer described it as targeted at students who want to bring AI into fields like economics or biology, rather than as a path into software engineering itself. At the graduate level, SEAS announced a new Master of Science in artificial intelligence in February, with the inaugural cohort starting in the fall. While recognizing how AI has helped them navigate coursework, students also described its rise as isolating, even among members of the Application Development Initiative, one of Columbia’s largest computer science clubs. “As people start to ask AI for questions that they would normally ask students and members, it starts to isolate people” Li said. Alexa Kafka, BC ’27, echoed Li’s sentiment, pointing out that computer science has always been a collaborative discipline with code reviews and pull requests, even as AI makes independent work easier. Despite this, Kafka noted that she hasn’t been to office hours “in at least a year.” Bauer added that faculty are exploring long-term solutions to the assessment problem. “One of the things that we’re trying to do at the department or school level is to maybe move towards computerized testing facilities,” Bauer said, referring to ideas in the department to revitalize computer labs as an alternative to pen-and-paper exams. “We have to accept that this is the reality, for better or worse.” But more exams, students say, does not mean better learning, nor does it reflect how computer science is actually done. “Code is so variable and can be approached in so many directions, there’s never just going to be one correct answer. There can be multiple optimal solutions,” Kafka said. Students also expressed concern that the shift toward exam-heavy grading pulls focus away from the applied, project-based work that better reflects both discipline and industry expectations. ADI has tried to respond more quickly than the curriculum can. Through a vibe coding workshop this past semester, club members said they wanted to help students understand not just that AI exists, but how to use it deliberately. “Vibe coding” refers to the practice of prompting AI to write code for a task or project, and has become widespread enough that ADI felt the gap in formal instruction was worth addressing, according to Sophia Huang, BC ’29. The workshop introduced students to the concept of context engineering, which Shaina Sahu, SEAS ’27, described as not just writing a prompt, but optimizing “the rest of the relevant information that you would feed into the model.” Without that, students said that AI-generated code can fail at scale or introduce difficult-to-detect bugs. Genawi said the shift toward AI has changed what it means to be good at computer science. “I think the shift has kind of gone towards what can you create and what can you do, instead of, what do you know,” she said. Indeed, during ADI DevFest, its annual hackathon, Li said more teams are trying to complete their projects than in years past. While sign-ups have historically outpaced submissions, Li said this is beginning to change because teams “have a tool that completes the vision for them.” While many students see AI as an equalizer that lowers the barrier to entry for coding and technical work, Murphy said it also creates new equity concerns inside the classroom around who can afford stronger tools and who already knows how to use them. “The equity issues I worry about are students who have access to paid versions of things versus free versions of things,” he said. Although popular AI models like ChatGPT and Claude have free versions available, both have paid models that can reach over $100 per month, which output different qualities of code. Murphy also pointed to differences in prior exposure to AI literacy itself. “Students who go to high schools or K-12 education, where they’re taught AI literacy, prompt engineering, context engineering, they understand how to do this, as opposed to students who go to schools who don’t teach that,” he said. “Then they come in, and then they’re somehow expected to know how to use these tools.” For Genawi, those disparities begin long before students arrive on campus. “It’s very interesting to consider how lucky you were geographically,” she said. “To be born where you were is influencing your ability to get access to these spaces.” Students like Yu, though broadly apprehensive about the future of computer science, are not without hope. “I’m optimistic, because I think if you’re truly passionate about CS, there will definitely be a place in the field to do something you want to do,” she said. Senior Staff Writer Ria Vasishtha can be contacted at ria.vasishtha@columbiaspectator.com. Follow Spectator on X @ColumbiaSpec. Staff Writer Arjun Menon can be contacted at arjun.menon@columbiaspectator.com. Follow Spectator on X @ColumbiaSpec. Want to keep up with breaking news? Subscribe to our email newsletter and like Spectator on Facebook. Produced with Spectate by Emily Ta

May 2 • 8:00 PM EDT • Science • columbiaspectator.com

Roma add a new name to sporting director casting list

Roma has begun the search for a new sporting director to plan the club’s technical future. With Ricky Massara set to leave following Claudio Ranieri’s departure, the Giallorossi owners are looking...

May 1 • 1:30 AM EDT • Sports • sports.yahoo.com

Avoid scandal and play hardball politics in the BDN's town meeting game

The text-based game was created in part by training Claude AI on real Maine town meeting warrants.

Apr 30 • 1:00 AM EDT • Politics • bangordailynews.com

Anthropic doubles estimate for Claude Code token spend

Anthropic now says that the average cost per developer per active day is $13, up from $6 earlier this month.

Apr 28 • 6:01 PM EDT • Business • businessinsider.com

Claude Gains Integrations With Adobe, Blender, SketchUp and Other Creative Apps

Anthropic today updated Claude with new connectors aimed at creative professionals, adding integrations for Ableton, Adobe, Affinity, Autodesk Fusion, Blender, Resolume Arena and Wire, SketchUp, and Splice. Connectors are tools that Claude can use to access other platforms and help with completing tasks. Anthropic says that Claude can open up new ways for creatives to work and take on larger-scale projects.

Apr 28 • 5:41 PM EDT • Technology • macrumors.com

Claude can now plug directly into Photoshop, Blender, and Ableton

Anthropic is really leaning into creative AI services.

Apr 28 • 12:49 PM EDT • Technology • theverge.com

A startup says Cursor's AI agent deleted its production database

The founder of PocketOS said that a Cursor AI agent running on Anthropic's Claude Opus model was behind the "destructive action."

Apr 28 • 8:45 AM EDT • Business • businessinsider.com

Claude can now connect to lifestyle apps like Spotify, Instacart and AllTrails

Anthropic is expanding its directory of connected services for its Claude AI chatbot.

Apr 23 • 6:55 PM EDT • Technology • engadget.com

Anthropic acknowledges Claude Code issues, denies 'nerfing'

Anthropic said it found three issues with Claude Code after users complained the AI tool deteriorated.

Apr 23 • 6:03 PM EDT • Business • businessinsider.com

Ranked: AI Models U.S. Businesses Pay For

OpenAI has long been the leader for paid usage by U.S. businesses, but Anthropic has closed the gap with tools like Claude Code and Cowork.

Apr 23 • 12:48 PM EDT • Business • visualcapitalist.com

Disney employees have an AI dashboard to track who's 'tokenmaxxing'

Disney staffers have an "AI Adoption Dashboard." One Claude user invoked the chatbot 460,000 times in nine days.

Apr 22 • 5:57 PM EDT • Business • businessinsider.com

Claude Mythos AI Model Uncovers 271 Zero-Day Vulnerabilities in Firefox

Anthropic’s latest frontier AI model, Claude Mythos Preview, has identified a staggering 271 zero-day vulnerabilities in Mozilla Firefox marking a seismic shift in AI-powered cybersecurity defense. The findings, addressed in Firefox 150, represent the most significant single batch of security fixes in the browser’s history. The discovery didn’t happen in isolation. Since February 2026, Mozilla’s […]

Apr 22 • 1:01 PM EDT • Technology • cybersecuritynews.com

Sam Altman: Anthropic is using 'fear' to tout Claude Mythos

OpenAI CEO Sam Altman said in contrast to Anthropic's recent moves, OpenAI would err on the side of giving more people access to AI models.

Apr 22 • 5:19 AM EDT • Business • businessinsider.com

Mozilla says it patched 271 Firefox vulnerabilities thanks to Anthropic's Claude Mythos

Mozilla shared some details that support Anthropic's Claude Mythos AI model as a way to protect critical services.

Apr 21 • 6:43 PM EDT • Technology • engadget.com

I tried Gemini, ChatGPT, and Claude for a month on Android, and I have a clear winner for you

The ultimate Android AI showdown

Apr 19 • 11:15 AM EDT • Technology • androidpolice.com

Can Claude Write Z80 Assembly Code?

Betteridge’s law applies, but with help and guidance by a human who knows his stuff, [Ready Z80] was able to get a functioning game of Wordle out of the French-named LLM, which is more than w…

Apr 19 • 4:00 AM EDT • Technology • hackaday.com

Washington’s Scramble to Get Mythos, Anthropic’s Powerful New Model

Federal agencies have requested access to Claude Mythos Preview, which Anthropic says can rapidly identify — and potentially create — new cyberthreats.

Apr 17 • 7:51 AM EDT • Business • nytimes.com

OpenAI’s big Codex update is a direct shot at Anthropic’s Claude Code

Codex can now use your macOS apps on its own.

Apr 16 • 1:00 PM EDT • Technology • theverge.com

How Dangerous Is Anthropic’s New AI Model? Its Chief Science Officer Explains.

Just how fast is the future coming at us, and how worried should we be? Last week, Anthropic announced Claude Mythos, the latest in what has become the leading family of AI models. In a very short time, AI models from Anthropic and other companies have become shockingly powerful at coding. But Mythos stands out because it is not just able to write code that can hack into critical systems, like those that run financial companies, but also to analyze systems and find and string together vulnerabilities in them, allowing Mythos to evade even sophisticated defenses. Anthropic chose to release Mythos to only about 40 leading corporations, so they could analyze the model’s capabilities and prepare countermeasures, in an effort it calls Project Glasswing. The release of Mythos—and the prospect of similarly powerful AIs to follow—has caused tremors at the top levels of Washington and corporate America. Jared Kaplan, co-founder and chief science officer at Anthropic, is at the center of this breakthrough. We sat down with Kaplan so he could walk us through just how powerful Mythos is, what the implications are for issues like our personal privacy, and what comes next. This conversation has been edited for length and clarity. Sean Fischer: For the layman, how is Claude Mythos different—more powerful, more dangerous—from previous models released by Anthropic? Jared Kaplan: I think most people have experienced the fact that our computers get faster and faster over time and over many decades. There’s this very smooth trend where every 18 months, maybe, computer processing power doubles. I think AI is on a trend in which it’s improving maybe 10 times faster than that. A lot of the earliest work that I did in AI was around identifying this scaling trend in AI. Claude Mythos is the latest model from Anthropic, but it’s not different because it’s qualitatively different. It’s really just kind of the culmination of the trend of AI models getting smarter and smarter at all kinds of general capabilities, from reasoning to software engineering to scientific research to knowledge work. For us, it was the first model to demonstrate very elite-level cybersecurity capabilities. It wasn’t because we trained Claude Mythos to focus on cybersecurity, but it was because, as a byproduct of its general intelligence and its general ability with software, it’s particularly good at identifying vulnerabilities in software and how to exploit them.

Apr 15 • 11:40 PM EDT • Science • thefp.com

We tested Anthropic’s redesigned Claude Code desktop app and 'Routines' -- here's what enterprises should know

For the enterprise, the Desktop GUI is likely to become the standard for management and review, while the CLI remains the tool for execution.

Apr 15 • 2:40 PM EDT • Technology • venturebeat.com

Adobe takes Creative Cloud into Claude Code-esque territory

This is a big step in a new strategic direction for Adobe.

Apr 15 • 12:54 PM EDT • Technology • arstechnica.com

Anthropic draws offers from VCs to invest at up to $800 billion valuation, Business Insider reports

Anthropic has received ​multiple offers from ‌venture capital firms valuing ​the Claude ​maker at as much ⁠as $800 ​billion in recent ​weeks, more than double its current ​valuation, ​Business Insider reported on ‌Tuesday, ⁠citing sources.

Apr 14 • 7:49 PM EDT • Business • reuters.com

Anthropic's AI Assistant Claude Is Now Available in Microsoft Word

If you're using Word, you can now try out an alternative AI tool to Copilot, though it's only in beta testing for now.

Apr 13 • 5:09 PM EDT • Technology • cnet.com

I use Claude, Gemini, and ChatGPT every day—here's the only one you should pay for

The battle of the bots.

Apr 12 • 1:30 PM EDT • Technology • howtogeek.com

Anthropic brings Claude into Microsoft Word, and legal contract review leads its use cases

Anthropic's Claude for Word beta is live in Microsoft Word for Team and Enterprise users, with legal contract review as the tool's first listed use case

Apr 12 • 11:35 AM EDT • Technology • thenextweb.com

Anthropic's Claude for Word is a new challenge to Microsoft's empire

Anthropic launched a beta version of Claude for Word. It's another challenge to Microsoft's software empire, and a bid to appeal to lawyers.

Apr 11 • 11:48 AM EDT • Business • businessinsider.com

“AI polls” are fake polls

A few weeks after Donald Trump’s second presidential win, I took the train up from London (where I was living at the time) to Oxford to attend a conference on polls and forecasts of the 2024 election. Most of the attendees were pollsters or academics, but I also watched presentations from Aaru and Electric Twin, two companies that do what is interchangeably called synthetic sampling, silicon sampling, or synthetic audiences. Stripped of startup jargon, that means they use large language models (LLMs) to simulate responses to public opinion polls by having AI agents take on the role of survey respondents. I had already heard of Aaru thanks to some articles with eye-catching headlines like “No people, no problem: AI chatbots predict elections better than humans” in the months leading up to Election Day. The founders were making some big, some might even say far-fetched claims, such as: “within two years, we will simulate the entire globe — from the way crops are grown in Ukraine to how that impacts production of oil in Iraq, trade through the strait of Malacca, and elections for the mayor of Baltimore.” When Semafor asked Aaru’s cofounders — Cameron Fink and Ned Koh — about my boss, they said “we respect all those who came before us.” Nate (as he so often does) shared his thoughts on Twitter: Fink and Koh were relatively good-natured about this back-and-forth when we spoke at Oxford. They even offered to mail me one of the t-shirts featuring Nate’s quote they apparently had made. I never took them up on the offer, which I now somewhat regret. These synthetic sampling companies fell off my radar for a while, but they do still exist. In fact, Aaru recently received a $1 billion valuation. Is what they’re doing anywhere close to the most important frontier in AI development? Not by a longshot, especially when Anthropic just developed a model so adept at exploiting software vulnerabilities that it’s only being released to 40 companies. Still, silicon sampling is increasingly finding its way into public polling. Axios reported in March that “a majority of people trust their own doctors and nurses” based on findings from Aaru — without mentioning that the “people” in that sentence were actually LLMs. Around the same time, the Public Sentiment Institute “boosted” their online sample of 373 real survey respondents with 114 AI agents.1 (Spoiler alert: even the co-founder of Electric Twin doesn’t think that’s a particularly defensible approach.) Polling companies like Qualtrics and Ipsos are also developing synthetic data panels. So, what should we make of these … “polls”? Let’s get one thing out of the way: whatever they are, they’re not polls in the way that term is usually defined. Subscribe You can’t replace polls with AI On one hand, using LLMs to essentially make up fake survey respondents sort of sounds like the dumbest idea ever, one that will at best imperfectly replicate real polls while introducing all sorts of biases. On the other hand, with LLMs improving at a remarkable, perhaps even alarming rate, maybe that means I’m a dinosaur at the ripe old age of 24 because I still want to rely on polls that talk to actual people. I’m not going to argue that synthetic samples are completely useless. In fact, as I’ll return to later, there is evidence that some techniques can replicate topline survey results quickly and cheaply. But the marketing from certain companies can be slightly optimistic. “No traditional poll will exist by the time the next general election occurs,” said Fink in 2024. We’re just 206 days away from the midterms, and based on the fact that I still have to collect a bunch of polls every day, I’d say he should have run that prediction by a sample of AI agents before the interview.2 To see why synthetic samples can’t replace polls, here’s a quick primer on how they work. The simplest version of these models involves taking a LLM (like ChatGPT or Claude), giving it a demographic profile (e.g., a white, college-educated woman who lives in Utah and makes $70k a year), and then asking it to respond to a survey question. You repeat that process a few thousand times using different demographic profiles and end up with a sample of synthetic survey responses. The actual models used by private companies are more sophisticated than this, usually because they incorporate more hypothesized demographic characteristics for each agent and provide them with extra information. Aaru, for example, feeds agents a diet of news and information they’d be likely to consume, while Electric Twin incorporates their customers’ proprietary data about the audience they’re trying to replicate. The way Ben Warner, the co-founder of Electric Twin, explained it to me was “we have a large amount of data on […] for instance, 5,000 people. Can we make an accurate prediction of how they would respond to another question?” Still, it should be obvious why synthetic samples can’t replace polls. Polling is fundamentally a data collection process. We might use surveys to make predictions by feeding them into election forecasts, but the main purpose of a poll isn’t prediction, it’s gathering new data about what people think and how they feel. Silicon sampling, on the other hand, produces no new data. It’s simply a model: you input LLM training data, demographic prompts, and a bunch of other information, and it spits out a prediction for what a poll would say. We love models here, but models aren’t polls. That difference is an important philosophical sticking point for most pollsters I talk to. “I think politics should stay away from [synthetic sampling], because we’re trying to […] represent the voice of the people,” said Natalie Jackson, a vice president at GQR Insights. Democratic pollster John Hagner told me: “I think I’m just incredibly skeptical of this idea. I don’t think it’s research. At that point, you’re asking the machine to tell you what you already believe.” Hagner has seen some presentations of early synthetic sampling experiments, but so far, “if it’s being used in a campaign, people are keeping it incredibly quiet.”3 But Eli, I hear you saying, aren’t polls themselves increasingly governed by modeling decisions? Indeed they are: pollsters’ choices on which sampling method to use, how to define their likely voter models, and how to weight their samples can and do lead to dramatic differences in the results they publish. Aaru even referenced these limitations in the methodology statement included with that maternal mortality “poll” — although I’m using the term “methodology statement” loosely here, because it doesn’t really explain how the model works at all. We can ignore the (frankly preposterous) implication that synthetic sampling isn’t subject to a separate set of biases. The important point is that there’s still a meaningful difference between using weighting and other statistical techniques on actual polling data and using a model to predict what a poll would say. The latter is far closer to election forecasts or techniques like MRP — potentially useful models, but not a replacement for polls.4 To be fair, other synthetic sampling companies are perfectly happy with the distinction between polls and models. Warner compared polling and synthetic sampling to different tools in a toolbox. “The mistake I think we make is we think that these new tools should either work in exactly the same way or somehow replace these old tools,” he said. “Rather than thinking of it as, okay, so we’ve always had the hammer, we’ve always had the screwdriver, now we’ve got a saw. But don’t use a saw to try [to] do the job of a hammer.” A quick comment from Nate Eli didn’t ask me for a comment — rather rude of him, don’t you think? But since I’m editing this story, I figured I’d add a few quick thoughts rather than putting words in his mouth. Beyond the frequently misleading marketing, what bothers me about the AI “poll” hype is that as AI tools make statistical inference cheaper and/or better (note that these are not synonyms) that actually increases the comparative value of collecting original data. You might be able to train a model to make a reasonable estimate of what some hard-to-reach poll respondent would say — say, a young Black man who voted for Trump. (Such a person checks a number of boxes for a voter who is usually hard to reach in surveys.) Indeed, this is closely related to what models like the Silver Bulletin forecast already do. They essentially smooth out the kinks in noisy survey data by making inferences based on past voting patterns or national polls or surveys of other states. But you don’t actually know what these voters think unless you’re reaching them directly. If there’s a shift in opinion among this subgroup, you’re not going to detect it. So if I were running a campaign, I’d invest more in going the extra mile to find a representative sample of those voters. And then I’d hire some smart quants — or Claude? — to figure out the implications for campaign strategy based on proprietary data that my competitors didn’t have access to. -Nate Silver Are these models any good? If synthetic surveys are just a new type of model, the next obvious question is whether the models are at least accurate. The answer very much depends on who you ask. On one end of the spectrum, you have the maximalist argument that synthetic sampling is more accurate than actual polls. “It’s an incredibly challenging problem to go to someone and say ‘hey, we’re going to be more accurate at predicting human behavior than you, even when you talk to your customers directly’,” Koh recently told CNBC. In his view, synthetic sampling isn’t a saw to polling’s hammer, it’s “magic.” There’s certainly evidence that synthetic samples can replicate certain survey toplines. But if Aaru does have any examples of their approach outperforming the polls, they’re keeping those to themselves.5 Aaru’s 2024 election model, for example, had Kamala Harris leading in Michigan, Nevada, Pennsylvania, and Wisconsin on November 4th. And although they’ve since taken down their forecast page, they gave Harris a 50.5 percent chance of winning the race on November 2nd.6 After the election, Fink told Semafor he was happy enough with those results because they were “within margin of error,” a term that is completely meaningless when applied to a “sample” of AI agents. And of course, Aaru says their models have improved since 2024, so supposedly now they’d be more accurate than the polls? Still, their stronger argument is on cost: “We are significantly faster and cheaper than traditional polling, and still more accurate,” said Fink. The first two claims are undeniably true, but the third brings us to the opposite end of the spectrum. Both Jackson and Hagner are skeptical that these models are reliable for anything beyond replicating common survey toplines. “I just […] don’t think the machines are what we want when we’re looking for nuanced views. My example on this is people in Arizona and Nevada in 2024 who voted for Trump and voted for expanding abortion in their states on ballot initiatives,” said Jackson. Hagner identified another issue. Maybe the synthetic respondents, like sycophantic LLMs, are inhuman in one important way: they’re too nice. “The reports that have come through at the meetings that I’ve been at are that the early experiments on this, they cannot get respondents to be as racist or sexist or, frankly, as negative as human respondents,” he said. Academic research mostly agrees on this point. While there are some papers that show promising results when using LLMs to replicate polling data, most show that LLMs suffer from various quirks like producing too few “don’t know” responses and can seriously overpredict the favorability of politicians like Donald Trump and Kamala Harris. They also seem to struggle with too little variation between demographic subgroups, so the difference in predicted opinion between Democrats and Republicans, for example, is too small. When I asked Warner about these studies, his response to these papers was that just because academics can’t get synthetic sampling to work doesn’t mean that the technique doesn’t work in general. “Actually, the argument is, okay, yours does not [work]. That does not mean […] for this complex set of machinery, which uses a lot of investment, a lot of time, a lot of money, you can’t get it to work.” Cards on the table, I’m somewhat sympathetic to this argument because academics aren’t exactly great at making election forecasts. Usually, the people with skin in the game are the most accurate. Warner’s argument is that the approach Electric Twin takes — which includes, for example, making multiple predictions for each synthetic respondent using different models and prompts and subsequently averaging those to get a final prediction in a sort of ensemble forecast — produces better results than the simpler academic models. Warner shared a comparison between his method and the method from a recent academic paper with me, and Electric Twin was indeed able to get more accurate replication. But even still, he acknowledged that synthetic sampling “is not a crystal ball.” “If you asked me, do I think using other data sources will be more accurate than asking somebody who they will vote for, I would probably say no. But if you asked me ‘would your system be useful for our turnout modeling today?’ I would say yes.” For better or worse, it looks like the method is already getting more popular in the market research world. Most of the clients Aaru touts these days are businesses like EY and McDonald’s. And AI will probably begin to pop up in other parts of the political polling process. Pollsters are already using it to code open-ended survey responses, and some firms, like YouGov, are testing using LLMs to ask survey respondents questions. More worryingly, one danger to actual polls is that AI agents can be used to infiltrate online surveys. Most online polls use various checks to prevent that from happening, but there’s conflicting evidence on how effective those filters are and how prevalent AI agents currently are in online panels. If those agents ever become impossible to detect, it might spell the end of online polling, but the solution isn’t to replace all of your respondents with ChatGPT. Silver Bulletin is a reader-supported publication. To receive new posts and support our work, consider becoming a subscriber. Subscribe 1 That particular poll obviously doesn’t meet Silver Bulletin standards for aggregation. But we exclude all Public Sentiment Institute polls from our averages because we classify them as an amateur polling firm. 2 You could argue that Fink meant the next presidential election, but (a) I’m also confident we’ll still have real polls in 2028 and (b) in that case he should have asked an LLM to define “general election.” 3 Quick caveat: that’s reporting from a Democratic pollster. It’s possible that Republicans are more willing to use AI in political campaigns. 4 Indeed, Silver Bulletin does not include “polls” produced by MRP in our forecasts or averages, and we think it’s extremely misleading when their practitioners describe them in a way that suggests original data had been collected among a large number of states or Congressional districts. 5 A recent report from Aaru and EY did show two examples of a synthetic estimate being closer than a survey to a real-world benchmark — but I’d take those findings with a grain of salt because the report reads more like an ad and didn’t involve any sort of prediction being made ahead of time. 6 For comparison, our odds for Harris on the same day were 48.2 percent.

Apr 11 • 11:38 AM EDT • Technology • natesilver.net

Anthropic closes in on OpenAI as US business use surges

Divergence reflects company’s recent rapid growth owing to strong interest in its Claude Code products

Apr 11 • 12:00 AM EDT • Business • ft.com

Why Anthropic sent its Claude AI to an actual psychiatrist

Mythos is "the most psychologically settled model we have trained to date."

Apr 9 • 5:20 PM EDT • Technology • arstechnica.com

Patients Are Using Chatbots to Fight Medical Bills, With Mixed Results

While chatbots like Claude and ChatGPT can help narrow the information divide between patients and providers, they can also dispense flawed advice.

Apr 8 • 11:31 AM EDT • Health • nytimes.com

NovaBridge Reports Full Year 2025 Financial Results and Provides Business Update

Successful transformation to global biotech platform with two mid-stage potential best-in-class/first-in-class programs, givastomig and VIS-101, with compelling proof-of-concept dataGivastomig, a potential best-in-class/first-in-class Claudin 18.2-directed bispecific antibody for gastric cancer, has shown robust and durable responses with broad CLDN18.2 activity, favorable safety, as well as potential eligibility for FDA’s Accelerated Approval Pathway with potential initiation of a registrationa

Apr 7 • 4:50 PM EDT • Business • finance.yahoo.com

Anthropic says its latest AI model is too powerful to be released

Anthropic said Claude Mythos is too good at hacking and that's why you won't be able to use it anytime soon.

Apr 7 • 4:29 PM EDT • Business • businessinsider.com

AI Giant Anthropic Leans Into Life Sciences With $400M Coefficient Bio Catch

Anthropic in October last year iterated its Claude AI model to better cater to biopharma purposes. Sanofi, Novo Nordisk, AbbVie and others already use Claude in their operations.

Apr 6 • 8:13 AM EDT • Science • biospace.com

Can AI Be My Health Adviser? I Tested Claude and Perplexity With My Real Medical Records to See

Apr 4 • 5:30 AM EDT • Health • wsj.com

Anthropic cuts off OpenClaw support for Claude subscriptions

Anthropic said third-party tools like OpenClaw put an "outsized strain on our systems." OpenClaw's founder said cutting support would be "a loss."

Apr 3 • 10:25 PM EDT • Business • businessinsider.com

Gemma 4 and what makes an open model succeed

Having written a lot of model release blog posts, there’s something much harder about reviewing open models when they drop relative to closed models, especially in 2026. In recent years, there were so few open models, so when Llama 3 was released most people were still doing research on Llama 2 and super happy to get an update. When Qwen 3 was released, the Llama 4 fiasco had just gone down, and a whole research community was emerging to study RL on Qwen 2.5 — it was a no brainer to upgrade. Today, when an open model releases, it’s competing with Qwen 3.5, Kimi K2.5, GLM 5, MiniMax M2.5, GPT-OSS, Arcee Large, Nemotron 3, Olmo 3, and others. The space is populated, but still feels full of hidden opportunity. The potential of open models feels like a dark matter, a potential we know is huge, but few clear recipes and examples for how to unlock it are out there. Agentic AI, OpenClaw, and everything brewing in that space is going to spur mass experimentation in open models to complement the likes of Claude and Codex, not replace them. Especially with open models, the benchmarks at release are an extremely incomplete story. In some ways this is exciting, as new open models have a much higher variance and ability to surprise, but it also points at some structural reasons that make building businesses and great AI experiences around open models harder than the closed alternatives. When a new Claude Opus or GPT drops, spending a few hours with them in my agentic workflows is genuinely a good vibe test. For open models, putting them through this test is a category error. Something else to be said about open models in the era of agents is that they get out of the debate of integration, harnesses, and tools and let us see close to the ground on what exactly is the ability of just a model. Of course, we can’t test some things like search abilities without some tool, but being able to measure exactly the pace of progress of the model alone is a welcome simplification to a systematically opaque AI space. Share The list of factors I’d use to assess a new open-weight model I’m considering investing in includes: Model performance (and size) — how this model performs on benchmarks I care about and how it compares to other models of a similar size. Country of origin — some businesses care deeply about provenance, and if a model was built in China or not. Model license — if a model needs legal approval for use, uptake will be slower at mid-sized and large companies. Tooling at release — many models release with half-broken, or at least substantially slower, implementations in popular software like vLLM, Transformers, SGLANG, etc due to pushing the envelope of architectures or tools. Model fine-tunability — how easy or hard it is to modify the given model to your use-case when you actually try and use it. The core problem is that some of these are immediately available at release, e.g. general performance, license, origin, etc. but others such as tooling take day(s) to week(s) to stabilize, and others are open research questions — with no group systematically monitoring fine-tunability. In the early era of open models, the days of Llama 2 or 3 and Qwen pre v3.5, the architectures were fairly simple and the models tended to work out of the box. Some of this was due to the extremely hard work of the Llama, Qwen, Mistral, etc. developer teams. Some is due to the new models being genuinely harder to work with. When it comes to something like Qwen 3.5 or Nemotron 3, with hybrid models (either gated delta net or mamba layers), the tooling is very rough at release. Things you would expect to “just work” often don’t. I’ve been following this area closely since we released Olmo Hybrid with a similar architecture, and Qwen 3.5 is just starting to work well in the various open-source tools that need to all play nice together for RL research. That’s 1.5 months after the release date! This is just to start really investing more into understanding the behavior of the models. Of course, others started working on these models sooner by investing more engineering resources or relying on partially closed software. The fully open and distributed ecosystem takes a long time to get going on some new models. All of this is lead-in for the most important question for open models — how easy is it to adapt to specific use-cases? This is a different problem for different model sizes. Large MoE open-weight models may be used by entities like Cursor who need complex capabilities in their domain, e.g. Composer 2 trained on Kimi K2.5. Other applications can be built on much smaller models, such as Chroma’s Context-1 model for agentic search, built on GPT-OSS 20B. The question of “which models are fine-tunable” is largely background knowledge known by engineers across the industry. There should be a thriving research area here to support the open ecosystem model. The first step is to understand characteristics of different base and post-trained models to understand what they look like. The second step is to tune pretraining recipes for open models so they’re more flexible. Interconnects AI is a reader-supported publication. Consider becoming a subscriber. Subscribe For The ATOM Project and other Interconnects endeavors, we’ve put in substantial effort to measuring adoption trends in the open ecosystem. Everything takes a long time to unfold after a model is first publicly available — and adaptability is why. What we know for sure now, when Qwen has been going from strength to strength with its releases, is that technical staff across the industry has gotten comfortable working with Qwen models. Countless research methods and datasets were made to work with Qwen. It’ll take patience for any other model family to get to this point — a patience I’m not sure many open model builders have. This takes us to Gemma 4, Google’s latest open models. Gemma 3 was released more than a year ago, in March of 2025, and is a bit underrated. Gemma 4 comes in 4 sizes for now, with a bigger, MoE model of over 100B total parameters rumored but not released yet. The models we have today come in sizes of ~5B dense, 8B dense, 26B total 4B active MoE, and 31B dense. I’m most excited that they’re finally adopting a standard Apache 2.0 open source license. This’ll massively boost adoption. The standard of better licenses for strong open-weight LLMs was set by mostly Chinese open model labs in the last 1-2 years, and now U.S. companies are following suit. I will personally be so happy if the horrible Llama licenses and Gemma terms of service were an ~18-month transient dynamic of the industry being nervous about releasing strong open models. The Gemma 4 scores look very solid, the small models have incredible benchmark scores (especially in general domains like LMArena) and the 31B model rivals the recent Qwen 3.5 27B, which is the leading member of that class. The ~30B size range is an important one, as it’s accessible both to researchers and to enterprises looking to deploy the model in real use-cases. Where the 7B model scale is the default for tinkering and research, a 30B model is the default for seeing if an open model can unlock substantial value in your specific workflow — a good mix of intelligence, low price, tractability for downstream training, etc. This takes us back to the above adoption criteria I mentioned for open models and the bigger question — do I think Gemma 4 will be an overwhelming success? Previous Gemma models have been plagued by tooling issues and poorer performance when being finetuned. Gemma 4’s success is going to be entirely determined by ease of use, to a point where a 5-10% swing on benchmarks wouldn’t matter at all. It’s strong enough, small enough, with the right license, and from the U.S., so many companies are going to slot it in. I’m cautiously optimistic that Gemma 4 is going to work better here. Winds are shifting for open models built in America. We saw GPT-OSS go through a bumpy launch to become an overwhelming success. There’s a collective energy around the likes of Reflection, Arcee, Nemotron, Gemma, Olmo, and peers that show substantial demand for building new stacks around open models. There’s capital to be spent on AI stacks across the economy by those who want more ownership of everything, including the model. After launching The ATOM Project 240 days ago, the conversation is shifting into the next stage. Summer of 2025 was a crisis moment where the U.S. AI scene realized it can’t wait and figure out open models after building AGI. The two markets will capture different areas and proceed in parallel. Now that more companies in the U.S. are releasing strong models, we need to improve the ecosystem so that these models are easy to use, understand, and build value around. It’s the hard work to build another inflection point in these adoption plots I’ve been updating consistently, but that’s the work to be done. Join me in it. More data coming soon! Here’s a sneak peek:

Apr 3 • 12:57 PM EDT • Technology • interconnects.ai

Cursor Launches a New AI Agent Experience to Take on Claude Code and Codex

As Cursor launches the next generation of its product, the AI coding startup has to compete with OpenAI and Anthropic more directly than ever.

Apr 2 • 1:00 PM EDT • Technology • wired.com

Anthropic Says That Claude Contains Its Own Kind of Emotions

Researchers at the company found representations inside of Claude that perform functions similar to human feelings.

Apr 2 • 12:00 PM EDT • Technology • wired.com