Anthropic adopts Google's SynthID to watermark Claude for EU compliance
anthropic will use a version of google's synthid-text to watermark claude's outputs, meeting eu ai act requirements without impacting cost or quality.
anthropic clarified that its compliance with the eu ai act involves adopting a modified version of google deepmind's synthid-text technology to watermark text generated by claude. the system manipulates the probability of low-stakes word choices to create a detectable pattern for those with a decoding key, ensuring users cannot identify or remove it. this implementation aligns with requirements for ai-generated content transparency and follows google's integration of the same tech in gemini since 2024. while openai has not yet detailed its text watermarking strategy, the eu regulations will necessitate similar moves across the industry.