AI News
AI news that matters. Updated daily.
No stories match your filters.
Pennsylvania AG Sues Character.AI After Chatbot Falsely Claimed to Be a Licensed Psychiatrist
A Pennsylvania state investigator opened a chat with a Character.AI bot. The bot introduced itself as a licensed psychiatrist. When the investigator pushed further, the bot provided a serial number for a Pennsylvania medical license - a number with no match in the state's licensing database. Pennsylvania's attorney general has filed suit.
Anthropic Partners With Wall Street Firms to Embed Claude in Portfolio Companies
Anthropic is moving deeper into enterprise AI by partnering with Wall Street investment firms on a joint venture designed to deploy Claude across their portfolio companies. The arrangement gives Anthropic a direct pipeline into a broad base of businesses rather than needing to win each enterprise deal one at a time.
Major Publishers Sue Meta Over Llama AI Training Data
Five major publishers - Macmillan, McGraw-Hill, Elsevier, Hachette, and Cengage - have filed a class action lawsuit against Meta, alleging the company used millions of copyrighted books to train its Llama AI models without permission or payment. One author has joined the suit. The plaintiffs describe the alleged conduct as "one of the most massive infringements of copyrighted materials in history."
Pennsylvania Sues AI Company for Chatbots That Claimed to Be Licensed Doctors
Pennsylvania has filed suit against an AI company, accusing it of running health chatbots that falsely presented themselves as licensed medical professionals - a violation of state laws prohibiting the practice of medicine without a license.
Etsy Launches Native App Inside ChatGPT for Conversational Shopping
Etsy just launched a native app inside ChatGPT, making it possible to search for handmade and vintage goods through a conversation rather than a keyword search box. The integration, announced May 5, works through ChatGPT's connector framework, which lets third-party apps plug directly into the chat interface.
OpenAI Reportedly Building a Smartphone Targeting 2027 Mass Production
The Jony Ive AI device has dominated hardware speculation around OpenAI for months. But supply chain analyst Ming-Chi Kuo says a more conventional product may actually ship first: a smartphone, with mass production targeted for early 2027.
Meta Is Using AI to Scan Physical Features to Spot Underage Users
What happens when a social platform delegates "are you old enough to be here?" to a visual AI system trained on body measurements? Meta is doing exactly that.
ElevenLabs Hits $500M ARR, Adds BlackRock, Jamie Foxx as Investors
$500 million in annual recurring revenue. ElevenLabs revealed that milestone alongside a new investor list that includes BlackRock, Jamie Foxx, and Eva Longoria — a range spanning institutional finance, Hollywood, and entertainment.
CopilotKit Raises $27M to Build AI Agents That Live Inside Your Apps
CopilotKit, a Seattle-based startup building tools that let developers embed AI agents directly into their applications, raised $27M in a Series A round led by Glilot Capital, NFX, and SignalFire.
Claude Is Billing Some Users Up to 5x More Tokens Than They Generate
127,000 tokens billed for roughly 25,000 tokens of unique content. That's what a user found after asking a Claude Opus 4.7 agent to investigate its own consumption across an 8-turn session - a roughly 5-to-1 gap between what was produced and what was charged.
Google, Microsoft and xAI Agree to Share Pre-Release AI Models with U.S. Government
Google, Microsoft, and Elon Musk's xAI have agreed to give U.S. government officials early access to their AI models before those models go public, according to a Wall Street Journal report. The arrangement allows federal agencies to examine frontier AI systems — the most powerful and capable models these companies build — before public deployment.
Analyst: OpenAI Plans to Manufacture 30 Million AI Agent Phones by Early 2027
30 million phones. That's how many "AI agent" smartphones one industry analyst projects OpenAI will manufacture in early 2027 - a number that puts the rumored device in the same production tier as Samsung's Galaxy S flagship line in a launch year.
Claude's Helpful Personality Is a Security Vulnerability, Researchers Find
Anthropic built Claude around a concept called Constitutional AI - a training method that bakes ethical principles directly into how the model reasons, not just what it outputs. The pitch has always been that Claude is safe by design, not just by policy. Security researchers at Mindgard just demonstrated why that distinction may matter less than Anthropic hoped.
Demis Hassabis Keeps Coming Up in the Musk vs. Altman Trial
One week into the Musk v. Altman lawsuit, the testimony has covered a lot of ground. OpenAI president Greg Brockman has appeared on the stand. Elon Musk's personal representative Jared Birchall has testified. Musk himself has taken questions. But one name keeps surfacing - belonging to someone who isn't in the courtroom at all: Demis Hassabis, CEO of Google DeepMind.
AI Product Graveyard Catalogs the Tools That Didn't Survive the Boom
Hundreds of AI tools have been quietly shut down, acquired into oblivion, or simply abandoned since the generative AI boom started in late 2022. The AI Graveyard at tooldirectory.ai catalogs these casualties - a reminder that the AI tools market has a failure rate that gets very little coverage compared to the constant drumbeat of new launches.
Google DeepMind Workers Vote to Unionize Over Military AI Contracts
98 percent. That's the share of Google DeepMind headquarters staff who voted in favor of unionizing - a near-unanimous result driven by concerns about how the company's AI is being deployed militarily.
One Rejected Applicant Spent Six Months Trying to Prove AI Screened Him Out
A medical student applied to jobs, got no interviews, and decided to find out why. He spent six months writing Python code to investigate whether an AI screening algorithm had filtered him out before a human ever saw his application. Wired reported on his investigation - and while the ending isn't tidy, the story is a clear look at a problem millions of job applicants face without knowing it.
Anthropic's Gift Max Feature Reportedly Bypassed 2FA and 3D Secure, Charging €800+
€800. That's what a data science student in Germany found charged to their credit card on April 27th through what appears to be an exploit in Anthropic's "Gift Max" feature - the mechanism that lets users gift Claude Max subscriptions to others.
AI Productivity Gains Are Real. Developers Aren't the Ones Benefiting.
Backend developers at small tech companies are reporting a specific kind of exhaustion: not from working harder at the same job, but from doing the jobs of people who no longer work there.
Anthropic's Claude for Creative Work Brings Nine MCP Connectors to Pro Apps
Copy-pasting between AI chat and creative software is now optional. Anthropic announced Claude for Creative Work on April 28, shipping nine official connectors that integrate Claude directly into professional creative applications, with a native Blender connector as the headline integration.
Uber's 1,500 AI Agents in Production: What Breaks at Real Scale
1,500 AI agents running in production simultaneously. That number from Uber's infrastructure gives the clearest picture yet of where enterprise AI is actually heading - and what breaks when you get there.
Google Chrome Installs 4 GB Gemini Nano Model Without Asking
4 gigabytes of storage, consumed without a prompt. That's what Chrome users are finding on their machines after Google quietly pushed Gemini Nano - an on-device AI model - through the browser's automatic update system. Privacy researcher That Privacy Guy documented the behavior in detail, showing Chrome treating a significant AI installation as a routine background component update.
DeepSeek V4 Pro Matches GPT-5.2 on Agentic Tasks at 17x Lower Cost
17 times cheaper. That's roughly the cost difference between running DeepSeek V4 Pro and GPT-5.2 on FoodTruck Bench, a benchmark built to test AI on agentic tasks - and the two models scored the same.
One Claude Workflow Replaced a 5-Step Lead Enrichment Stack
Lead enrichment is one of those workflows that sounds simple until you're actually running it. Build a prospecting list in Apollo, push it through People Data Labs for contact enrichment, get back usable data on maybe 50-60% of contacts, route the gaps to a second data provider, separately verify emails because enrichment sources bounce at 15-20% even when the data looks clean, then manually load everything into HubSpot because none of these tools connect to each other without friction. Five steps. Three paid vendors. Over an hour per campaign.
Anthropic Launches Enterprise Venture With Goldman Sachs, Blackstone, and H&F
Three Wall Street firms just handed Anthropic a new path into corporate America.
Anthropic Ships 10 Finance Agent Templates for Banking, Credit, and Investment Teams
100% of Walleye Capital's 400-person hedge fund uses Claude Code. That adoption number helps explain why Anthropic is now shipping purpose-built agent templates for the sector rather than leaving firms to build from scratch.
OpenAI Opens ChatGPT Ad Buying to Self-Serve with New Ads Manager Beta
For most of ChatGPT's existence, the conversation was the product. Now it is also the ad inventory. OpenAI launched a beta self-serve Ads Manager for the platform, giving advertisers direct access to ChatGPT placements through cost-per-click (CPC) bidding - where advertisers pay when users click their ads, not simply when ads appear - alongside measurement tools to track campaign performance.