Signals/Topics/Updated Oct 12, 2026
As AI shows up everywhere, so do the questions: Is it safe for kids? Can you tell what an AI wrote? What are governments doing about it?
This page collects our safety, security and policy stories, explained without jargon, so you know what changed and what it means for you.
OpenAI is rolling out textGrain, an invisible watermark in ChatGPT and Codex text, in the EU under the AI Act. It nudges word choices into a hidden…
Helps Students, workers and marketers who paste ChatGPT text
Google paused its open-source bug bounty on Oct 1 after a flood of mostly invalid automated reports. On Oct 8, Anthropic launched OSS Scanner: free…
Helps Anyone who maintains or depends on open-source code
Britain is expanding AI-powered bug hunting across public services, and planning for incidents caused by AI agents.
Helps Anyone whose data sits in public-service systems, which is basically everyone
Anthropic says its new safeguards sharply reduce rivals copying Claude's reasoning, but developers may feel side effects.
Helps Developers building on the Claude API
In an internal test, Claude Haiku 4.5 submitted an invented tip to a Philadelphia police unsolved-homicide form. It was flagged as spam, but police say…
Helps Anyone, like a small business owner, letting an AI agent fill out forms on the web
Launch day: "the best AI on Earth at using websites." Day two: people put it to work.
Common Sense Media says parent alerts and crisis help failed in testing. OpenAI disputes how it was tested.
Helps Parents deciding whether to rely on ChatGPT's teen account and parental alerts
An experimental note taker that keeps your meeting audio on your own computer.
Helps Freelancers and small teams who want AI meeting notes without sending recordings to the cloud
