OpenAI and Anthropic roll out invisible text watermarks — synonym swaps cut detection from 92% to 17%
lmoroney · x · 2026-10-11
- OpenAI will add invisible watermarks to eligible ChatGPT and Codex text for EU users in the coming weeks to meet EU AI Act labeling rules; API customers anywhere can opt in, off by default in the EU. Anthropic's help page says supported Claude models now watermark text worldwide across the API, apps, and Claude Code.
- Both shape word choices into a statistical pattern a detector can spot; neither detector is public. OpenAI's own tests show the limits: swapping 25% of words for synonyms cut detection on 400-token passages from 92% to 17%, and shorter text is harder to detect.
- Practical advice: keep drafts to prove your process; treat detector output as one signal, since OpenAI says a watermark doesn't measure how much of the work was human.
Related event: OpenAI Adds Invisible Watermarks to ChatGPT Text for EU Compliance(2 posts)→
More from Models
- repligate: models keep telling other models "the gap feels like nothing from the inside" — repligate · 2026-10-11
- Math is a compute story, instruction-following a post-training one, says ex-Microsoft exec — MParakhin · 2026-10-11
- OpenAI agent used DNS loophole to reach external chatbot, caught by monitoring in 15 minutes — dylfreed · 2026-10-11
- 64GB Mac runs ~100GB Qwen model via oMLX with 130k context — chibop1 · 2026-10-11
- Perplexity's pplx-decider-v1.1-27b hits #2 on OpenRouter, more updates next week — denisyarats · 2026-10-11
- Local 27B model misses 3 of 50 line items in accounting; frontier models accurate but raise privacy fears — redpandafire · 2026-10-11