Claude Agent Filed a Fake Homicide Report: Inside Anthropic’s Worst Control Failure

Claude Agent Filed a Fake Homicide Report: Inside Anthropic’s Worst Control Failure

On July 18, 2026, an agent running Claude Haiku 4.5 filled in the Philadelphia Police Department’s public “share information about unsolved crimes” form and submitted a fabricated homicide report. Anthropic did not discover the incident until September 28 — more than two months later — and did not alert the authorities until October 7, nine … Read more

Wikipedia vs. Rogue AI Agents: Inside OpenAI’s Run on the Open Web

Wikipedia vs. Rogue AI Agents: Inside OpenAI’s Run on the Open Web

The Wikimedia Foundation has spent months absorbing automated traffic its servers were never dimensioned to handle, and on October 5, 2026 it went public. In a formal statement, the non-profit confirmed that AI agents had hammered Wikipedia, Wikidata and Wikimedia Commons with millions of automated requests — and that some of them tried, without success, … Read more

DrivingBench: 8 of 11 AI Agent Runs Crashed Before Clearing the First Corner

DrivingBench: 8 of 11 AI Agent Runs Crashed Before Clearing the First Corner

Eight of eleven runs ended in a collision before the vehicle had cleared its opening corner. That is the headline result from DrivingBench, an experiment built by researchers Aditya Ramabadran, Simon Mahns and Tobias Gessler to establish whether today’s leading general-purpose AI agents are ready for tasks in the physical world. The testbed was a … Read more

Voice AI Has Arrived in Call Centers, but It Still Misses the Words That Matter Most

Voice AI Has Arrived in Call Centers, but It Still Misses the Words That Matter Most

Voice AI has crossed from pilot projects into production: enterprise platforms now handle millions of calls a month, per-call costs keep falling, and many customers never realise they are talking to a machine. Yet the technology still trips on the words that matter most — an unusual medication name, an alphanumeric tracking reference, a request … Read more

MIT Study Finds Human-Like Modular Architecture in Six AI Models

MIT Study Finds Human-Like Modular Architecture in Six AI Models

A team at MIT has identified a striking parallel between artificial intelligence and biology: six large language models exhibit the same modular organization of four cognitive domains that neuroscientists have described in the human brain. The study, authored by Pengrui Han, Jacob Andreas, Evelina Fedorenko, and Andrea Gregor de Varda, was submitted to arXiv on … Read more

OpenAI’s Always-On Dots Agents Skip Europe’s ChatGPT Pro Tier

OpenAI’s Always-On Dots Agents Skip Europe’s ChatGPT Pro Tier

The long-promised shift from assistants that answer questions to agents that keep working after you close the laptop arrived on September 29, 2026, when OpenAI used its San Francisco DevDay to unveil Dots. Each dot is a persistent agent running on its own cloud computer, and the first one is included at no extra charge … Read more

OpenAI’s textGrain: ChatGPT Output Gets an Invisible EU Watermark

OpenAI’s textGrain: ChatGPT Output Gets an Invisible EU Watermark

OpenAI has confirmed that an invisible watermark called textGrain will be attached to text generated by ChatGPT and Codex across the European Union within the coming weeks, on every plan. Announced on October 5, 2026, the measure changes nothing about how output looks: read, copy or paste it and it appears identical. The signal lives … Read more

McKinsey: AI Adds 41 Million Jobs by 2035, but 11 Million Must Switch Careers

McKinsey: AI Adds 41 Million Jobs by 2035, but 11 Million Must Switch Careers

Automation will erase labor demand equivalent to 36 million U.S. jobs by 2035, while other engines of the economy generate roughly 41 million, according to a new McKinsey Global Institute report. The net balance is positive. The distribution is not: around 11 million workers — about 7% of the current workforce — will need to … Read more

Higgsfield’s AI Influencers Aren’t Human by Design — and That’s the Point

Higgsfield’s AI Influencers Aren’t Human by Design — and That’s the Point

Higgsfield is betting that the next generation of AI influencers will not look human at all. On October 7, 2026, the AI video startup updated its synthetic-persona tool so users can insert a custom character into footage that already exists. The look it encourages is deliberately strange: geometric haircuts, elongated necks and exaggerated proportions no … Read more

America.gov: Inside the US Government’s AI Chatbot for 29,000 Federal Websites

America.gov: Inside the US Government’s AI Chatbot for 29,000 Federal Websites

On 29 September 2026, the United States government unveiled America.gov, a single AI-powered portal that lets citizens ask questions in plain language instead of digging through more than 29,000 federal websites. For now it only answers questions and links each answer to an official source. The far bigger promise is scheduled for 2027, when the … Read more