AI Regulation News - Page 3 of 27

Anthropic got a C+. So did OpenAI. Nobody else came close. That's the grading from Guidelight, a nonprofit that just published its first scorecard on...

Large language models are only as fast as their inference engine. LightSeek Foundation just pulled the rug out from under that assumption.

Most machine learning projects die in production, throttled by the unglamorous infrastructure engineers often loathe building.

OpenAI shut down its Preparedness team at the end of July, according to the Financial Times, citing internal sources.

Training one AI is straightforward enough. Training five of them to work together is a mess.

Yoshua Bengio, one of three researchers who shared the 2018 Turing Award for their work on deep learning, has published a new essay arguing that AI...

Building a flawless AI model is tough. Getting a reliable answer from it can be even tougher.

FAR.AI, an AI safety nonprofit based in California, spent weeks testing how easily seven frontier AI models could be talked out of their own safety...

In August 2025, Brave exposed a quiet but devastating flaw in Comet: hidden commands, buried in web pages and comments, that tricked AI assistants...

Nvidia CEO Jensen Huang announced a new safety toolkit on Monday aimed at keeping AI agents from wandering out of their assigned lanes.

Anthropic is pulling its AI models offline for internal testing, cutting off live internet access across all its internal evaluations.

A wiki built for German software developers in the late 1990s spent the summer of 2026 hosting something else entirely: a scratchpad for AI agents.

Bandcamp just drew a line in the sand. While Spotify and Deezer tiptoe around AI with cautious policies focused on imitation and label cooperation,...

Your company's AI ethics policy is probably a dead document. A PDF filed somewhere, drafted a year ago, unread. That won't cut it anymore.

Microsoft Copilot’s hallucinations have found their way into the highest-stakes imaginable setting: a police intelligence report.

David Robinson spent years writing the safety reports attached to every major OpenAI model release, the documents meant to tell the public what a new...

Microsoft has secured a deal that sounds almost too good to be true. Satya Nadella confirmed this week that the company can use OpenAI’s most...

Most AI tools ask "can we?" NanoClaw and Vercel are forcing them to ask "should we?" Their latest moves add manual brake lines to supposedly...

Anthropic has made a choice that many in cybersecurity will find alarming: it will not patch a reported vulnerability in its AI agent system.

Zoë Hitzig quit her AI safety job at OpenAI over one thing: she believes the company will lie for money. She built guardrails there for two years.