Anthropic essentially bans OpenClaw from Claude by making subscribers pay extra
Using OpenClaw with Claude AI is about to get a lot more expensive, thanks to Anthropic's new policy changes. Beginning April 4th at 3PM ET, users will "no longer be able to use your Claude subscription limits for third-party harnesses including OpenClaw,". Instead, if users want to use OpenClaw with Claude, they'll have to use a "pay-as-you-go option" that will be billed separate from their Claude subscription. With OpenClaw creator Peter Steinberger now employed by OpenAI, Anthropic may also be encouraging subscribers to use more of its own tools, like Claude Cowork, instead.
Chatbots are now prescribing psychiatric drugs
Utah is allowing an AI system to prescribe psychiatric drugs without a doctor. It's only the second time the state – and the country – has delegated this kind of clinical authority to AI. State officials say it could bring costs down and ease care shortages, but physicians warn the system is opaque, risky, and unlikely to expand mental health care to those who need it. The one-year pilot, announced last week, will allow Legion Health's AI chatbot to renew certain prescriptions for psychiatric medications, in some cases. The San Francisco startup promises Utah-based patients "fast, simple refills" through a $19-a-month subscription.
Codex now offers more flexible pricing for teams
April 2, 2026 We’re making it easier to just build things. Starting today, teams on ChatGPT Business and Enterprise can add Codex-only seats to their workspaces with pay-as-you-go pricing, giving full access to Codex without a fixed seat fee. Now, small groups can begin pilots, prove value in a few critical workflows, and easily expand from there. Teams that need broad ChatGPT access can continue using standard ChatGPT Business seats that do include Codex usage limits. To support your adoption, eligible ChatGPT Business workspaces can receive $100 in credits for each new Codex-only team member that joins and starts using Codex, up to $500 per team, for a limited time.
The gig workers who are training humanoid robots at home
As companies like Tesla, Figure AI, and Agility Robotics race to build humanoids—robots designed to resemble and move like humans in factories and homes—videos recorded by gig workers like Zeus are becoming the hottest new way to train them. Micro1 has hired thousands of contract workers in more than 50 countries, including India, Nigeria, and Argentina, where swathes of tech-savvy young people are looking for jobs. For robots to work in factories and serve as housekeepers, real-world data, however time-consuming and expensive to collect, may be what we need. Investors are pouring money feverishly into solving this challenge, spending over $6 billion on humanoid robots in 2025.
Gradient Labs gives every bank customer an AI account manager
April 1, 2026 Gradient Labs uses GPT‑4.1 and GPT‑5.4 mini and nano to run complex financial support workflows with high accuracy and low latency. Results 10x Revenue growth Results 98% Customer satisfaction with AI agent experience Results +11% Higher accuracy with GPT-4.1 vs. next-best provider In banking, resolving a customer issue is rarely simple. Cases like fraud or blocked payments require strict adherence to complex procedures across multiple teams. Gradient Labs(opens in a new window) is built to handle this complexity. The London-based company is building AI agents that give every bank customer the experience of a dedicated account manager. The team built a hybrid architecture that uses OpenAI models for reasoning-intensive steps and smaller models for faster, deterministic tasks, with routing that adapts based on complexity and latency constraints.
There are more AI health tools than ever—but how well do they work?
Earlier this month, Microsoft launched Copilot Health, a new space within its Copilot app where users will be able to connect their medical records and ask specific questions about their health. A couple of days earlier, Amazon had announced that Health AI, an LLM-based tool previously restricted to members of its One Medical service, would now be widely available. These products join the ranks of ChatGPT Health, which OpenAI released back in January, and Anthropic’s Claude, which can access user health records if granted permission. Health AI for the masses is officially a trend. There’s a clear demand for chatbots that provide health advice, given how hard it is for many people to access it through existing medical systems.
Why Are Large Language Models so Terrible at Video Games?
Large language models (LLMs) have improved so quickly that the benchmarks themselves have evolved, adding more complex problems in an effort to challenge the latest models. While a few have managed to beat a few games (for example, Gemini 2.5 Pro beat Pokemon Blue in May of 2025), these exceptions prove the rule. Julian Togelius, the director of New York University’s Game Innovation Lab and co-founder of AI game testing company Modl.ai, explored the implications of LLMs’ limitations in video games in a recent paper. He spoke with IEEE Spectrum about what this lack of video games skills can tell us about the broader state of AI in 2026.
STADLER reshapes knowledge work at a 230-year-old company
March 27, 2026 Embedding ChatGPT across 650 employees to turn hours of knowledge work into minutes—scaling speed, quality, and decision-making company-wide. Results 125+ custom GPTs created Results 30-40% time savings on common knowledge tasks Results 2.5x faster time to first draft on average Results >85% daily active usage STADLER is a family-owned company with more than 230 years of history, specializing in automated waste sorting plants for the global recycling industry. Under the leadership of Co-CEO Julia Stadler, the company has taken a forward-looking approach to modernization—embedding AI into everyday work as a core productivity layer. Since 2023, STADLER has pursued a clear principle: every employee working on a computer should use AI to improve speed, quality, and collaboration.
The Download: the internet’s best weather app, and why people freeze their brains
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. Stephen Coles. Coles, a researcher who studied aging, was interested in cryonics—the long-term storage of human bodies and brains in the hope that they might one day be brought . It’s a hope shared by many. Over the past few years, I’ve spoken to people who run cryonics facilities, study cryopreservation, or just want to be cryogenically stored. (404 Media) + Here’s what we’re getting wrong about AI’s truth crisis. (MIT Technology Review) 8 OpenAI’s ad pilot generated $100 million in under 2 months More than 600 advertisers are working on the trial.
Google is making it easier to import another AI’s memory into Gemini
Google is rolling out two new Gemini tools on Thursday which could attract users who are thinking about switching their preferred AI chatbot but don't want to "retrain" a new AI. Gemini's new "Import Memory" and "Import Chat History" features help users quickly copy over everything their current AI already knows about them with a couple of pre-written prompts. To use the "Import Memory" tool, users copy and paste a suggested prompt from Gemini into their previous AI, then paste the output from the previous AI into Gemini, which should get Gemini caught up on their preferences. The "Import Chat History" feature has users request an export o … Read the full story at The Verge.
Apple will reportedly allow other AI chatbots to plug into Siri
Apple's iOS 27 update will allow users to choose the AI chatbot they want to link with Siri. That's according to a report from Bloomberg's Mark Gurman, who says third-party chatbots downloaded from the App Store, like Google's Gemini or Anthropic's Claude, will be able to fetch replies for Siri – similar to how the voice assistant already works with OpenAI's ChatGPT. The new system, reportedly called "Extensions," will give users the ability to enable or disable the chatbots they want to connect with Siri on the iPhone, iPad, and Mac, Bloomberg reports. The upcoming integrations will also work with the standalone app Apple reportedly plans … Read the full story at The Verge.
Gemini 3.1 Flash Live: Making audio AI more natural and reliable
Mar 26, 2026 Our latest voice model has improved precision and lower latency to make voice interactions more fluid, natural and precise. Gemini 3.1 Flash Live is Google's highest-quality audio model, designed for natural and reliable real-time dialogue. Developers can access it through the Gemini Live API in Google AI Studio, while enterprises can use it for customer experience. Everyone can experience it via Search Live and Gemini Live, which now supports over 200 countries. Today, we’re advancing Gemini’s real-time dialogue capabilities with Gemini 3.1 Flash Live, our highest-quality audio and voice model yet. Illustrative demonstration built with Gemini 3.1 Pro, powered by Gemini 3.1 Flash Live. 3.1 Flash Live lets you use your voice to vibe code and quickly iterate.
Search Live is expanding globally
Mar 26, 2026 When you go Live with Search, you can have an interactive, multimodal conversation in AI Mode. Today, we’re expanding Search Live globally to all languages and locations where AI Mode is available. Search Live is designed for those moments when you need real-time help, and typing out a query just won’t cut it. To go Live with Search, simply open the Google app on Android or iOS and tap the Live icon under the Search bar. You can also access Search Live if you’re already pointing your camera with Google Lens — just tap the Live option at the bottom of the screen to have a real-time, back-and-forth conversation about what you see in the real world.
OpenAI shelves erotic chatbot ‘indefinitely’
OpenAI has paused plans to release a sexualized "adult mode" for ChatGPT, in its latest move to refocus on the company's core products. the erotic chatbot has been shelved "indefinitely" after facing pushback from employees and investors due to the problematic and harmful effects sexualized AI content can have on society. This decision comes in the wake of OpenAI also discontinuing its text-to-video AI platform Sora, citing "internal discussion about our broader research priorities." It's the latest side quest to be dropped by the company after CEO Sam Altman declared a "code red" in December, suggestin … Read the full story at The Verge.
Senate Democrats are trying to ‘codify’ Anthropic’s red lines on autonomous weapons and mass surveillance
Anthropic's fight with the Pentagon is expanding to Congress. Adam Schiff (D-CA) is working on a new bill to "codify" Anthropic's red lines and ensure humans make the ultimate decisions in questions of life and death, and Sen. Elissa Slotkin (D-MI) recently introduced a bill to limit the Defense Department's ability to use AI for mass surveillance of Americans. The Trump administration blacklisted Anthropic earlier this month after it set limits on how the military could use its AI models, designating it a supply-chain risk. Anthropic has filed suit, accusing the government of violating its constitutional rights. It's insisted that the … Read the full story at The Verge.
Anthropic’s Claude Code gets ‘safer’ auto mode
Anthropic has launched an "auto mode" for Claude Code, a new tool that lets AI make permissions-level decisions on users' behalf. The company says the feature offers vibe coders a safer alternative between constant handholding or giving the model dangerous levels of autonomy. Claude Code is capable of acting independently on users' behalf, a useful but risky feature as it can also do things users don't want, like deleting files, sending out sensitive data, and executing malicious code or hidden instructions. Auto mode is designed to prevent this, flagging and blocking potentially risky actions before they run and offering the agent a chan … Read the full story at The Verge.
The AI Hype Index: AI goes to war
Anthropic and the Pentagon feuded over how to weaponize Anthropic’s AI model Claude; then OpenAI swept the Pentagon off its feet with an “opportunistic and sloppy” deal. People marched through London in the biggest protest against AI to date. If you’re keeping score, Anthropic—the company founded to be ethical—is now turbocharging US strikes on Iran. On the lighter side, AI agents are now going viral online. OpenAI hired the creator of OpenClaw, a popular AI agent. Meta snapped up Moltbook, where AI agents seem to ponder their own existence and invent new religions like Crustafarianism. And on RentAHuman, bots are hiring people to deliver CBD gummies.
OpenAI just gave up Sora and its billion-dollar Disney deal
On Tuesday afternoon, OpenAI announced "We're saying goodbye to Sora," the video generation tool that it launched at the end of 2024, and centered in a massive licensing deal with Disney only a few months ago. The Wall Street Journal reported the move earlier, saying that OpenAI boss Sam Altman had informed staff that both the TikTok-like Sora app and API access for developers would be discontinued, with no plans to roll the feature into ChatGPT as had previously been rumored. as a result, the deal Disney announced in December, saying it would invest $1 billion in OpenAI, license its characters for use … Read the full story at The Verge.
ChatGPT and Gemini are fighting to be the AI bot that sells you stuff
The AI-powered shopping rivalry is heating up as Google and OpenAI launch new features to help you buy things while interacting with their chatbots. Now, Google is teaming up with Gap Inc to allow its Gemini AI assistant to purchase clothes on your behalf from any of its stores, which include Gap, Old Navy, Banana Republic, and Athleta, while OpenAI launched an updated shopping interface in ChatGPT. With Google's new partnership, you may see products from one of Gap's brands if you're asking Gemini for shopping suggestions, along with an option to purchase the item without leaving the chatbot. Similar to Walmart and Target, which also let u … Read the full story at The Verge.
Automating complex finance workflows with multimodal AI
Finance leaders are automating their complex workflows by actively adopting powerful new multimodal AI frameworks. Historically, standard optical character recognition systems failed to accurately digitise complex layouts, frequently converting multi-column files, pictures, and layered datasets into an unreadable mess of plain text. Platforms such as LlamaParse connect older text recognition methods with vision-based parsing. Specialised tools aid language models by adding initial data preparation and tailored reading commands, helping structure complex elements such as large tables. Within standard testing environments, this approach demonstrates roughly a 13-15 percent improvement compared to processing raw documents directly. Building scalable multimodal AI pipelines for finance workflows Successful implementation requires specific architectural choices to balance accuracy and cost.
Anthropic’s Claude Code and Cowork can control your computer
The new feature can be used to automatically open files, use web browsers and apps, and run dev tools "with no setup required," even when you're away from your computer, according to Anthropic's announcement. These new capabilities are available as a research preview for Claude Pro and Max subscribers, and computer usage is limited to macOS "for now,". The feature builds on autonomous capabilities that were introduced on Claude's 3.5 Sonnet AI model in 2024, but now brings those to the chatbot's C … Read the full story at The Verge.
Powering product discovery in ChatGPT
ChatGPT introduces richer, visually immersive shopping powered by the Agentic Commerce Protocol, enabling product discovery, side-by-side comparisons, and merchant integration.
The hardest question to answer about AI-fueled delusions
I was originally going to write this week’s newsletter about AI and Iran, particularly the news we broke last Tuesday that the Pentagon is making plans for AI companies to train on classified data. AI models have already been used to answer questions in classified settings but don’t currently learn from the data they see. That’s expected to change, I reported, and new security risks will result. Read that story for more. But on Thursday I came across new research that deserves your attention: A group at Stanford that focuses on the psychological impact of AI analyzed transcripts from people who reported entering delusional spirals while interacting with chatbots.
The gen AI Kool-Aid tastes like eugenics
Like many people, director Valerie Veatch was intrigued when OpenAI first released its Sora text-to-video generative AI model to the public in 2024. Though she didn't fully understand the technology, she was curious about what it could do, and she saw that other artists were building online communities to share their new AI creations. The hope of connecting with people drew Veatch into the AI space, but once she was there, she was shocked to see how often the technology would generate images dripping with racism and sexism. Veatch was even more unsettled by the way her new AI-enthusiast peers did not seem to care that the machine they ralli … Read the full story at The Verge.