Anthropic is expanding its text watermarking system to four additional Claude models, including Claude Mythos 5, Opus 4.7, Opus 4.6, and Sonnet 4.6, to comply with the EU AI Act. While newer models already have this feature, the older ones are being updated to embed an invisible statistical pattern in word choices. However, a loophole exists on Claude Opus 4.6 and Sonnet 4.6, where altering a single sampling parameter can bypass the watermark, a fact Anthropic has warned users about. AI
IMPACT Increases AI-generated content traceability for regulatory compliance, though a parameter loophole on older models limits its current airtightness.
RANK_REASON Expansion of a key AI transparency feature across multiple models, with a noted technical loophole.
- Anthropic
- Claude Mythos 5
- Claude Opus 4.6
- Claude Opus 4.7
- Claude Opus 5.5
- Claude Sonnet 4.6
- EU AI Act
AI-generated summary · Google Gemini · from 5 sources. How we write summaries →