Coders Claim Workarounds to Claude's Invisible Watermarks Within Hours
Anthropic announced last week it would embed invisible watermarks in AI-generated content from its Claude model to comply with new European Union rules. Within hours of the announcement, coders were publicly touting overrides and workarounds online, asserting that the watermarking system could be bypassed. The rapid response highlights ongoing challenges in enforcing provenance and traceability standards for AI-generated media under evolving regulatory frameworks.
Key facts
- Anthropic announced last week it would include invisible watermarks in AI-generated content.
- The watermarking move is intended to comply with new EU rules.
- Coders began touting workarounds online within hours of the announcement.
- The overrides claim the watermarks can be bypassed.
- The development raises questions about the enforceability of AI content provenance standards.
Entities
Institutions
- Anthropic
- European Union
Sources
- Wired AI —