Radar / AI security / OpenAI textGrain watermarking for ChatGPT and…
OpenAI textGrain watermarking for ChatGPT and Codex in the EU
OpenAI will roll out its invisible textGrain watermarking system to ChatGPT/Codex users in the EU over the coming weeks to meet the EU AI Act's machine-readable AI-content-identification requirement, following Anthropic's similar August release.
Why it matters
First large-scale deployment of invisible text watermarking by a frontier lab; OpenAI disclosed the detector's real limits — editing just 10% of words drops detection accuracy from ~92% to ~66%, and it struggles with short text, math answers, and translated text — and is initially limiting detector access to approved researchers.
What you could build with it
A newsroom or exam-proctoring startup could use watermark detection as one signal in a content-provenance pipeline, flagging suspected AI-generated passages for human review before publication rather than relying on detector scores alone.
Does it hold up?
Too early to judge — the rollout starts in the coming weeks; OpenAI itself says the detector struggles with short passages, math answers, and translated text, and has only granted detector access to approved researchers, so no independent validation exists yet.
Learn more
First spotted on article: source.
More AI security
OpenAI, Anthropic and Google DeepMind jointly unveil cyber-focused safety models and safeguardsThe three rival labs disclosed weeks of behind-the-scenes coordination and jointly released a set of cyber-focused AI…security · JEV 0.73Cloudflare security-audit-skill: multi-phase security audits for coding agentsA coding-agent skill that runs multi-phase security audits with independently verified, machine-checkable findings.security · JEV 0.63LiveNerf: a pre-registered benchmark for post-release model driftOpen-source, pre-registered 30-day benchmark that detects whether Claude Opus 5.5 quietly gets worse after launch.security · JEV 0.61Sandlock 0.8.8: deferred commit for agent sandboxesThe process-based Linux AI sandbox (no container, no VM) ships deferred commit: every run returns a changeset of what…security · JEV 0.58OpenAI disrupts Moonshot-linked 'adversarial distillation' campaignOpenAI disclosed it shut down a coordinated campaign by 15,000+ users to extract hidden model reasoning, attributing a…security · JEV 0.58ClawSecure: free security scanner for OpenClaw AI agent skillsFree scanner that audits OpenClaw agent skills for vulnerabilities; the maker's audit of 2,890+ OpenClaw skills found…security · JEV 0.5
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.