Tech Meridian ← LIVE FEED
RU

RESEARCH · RESEARCH · #613

Research finds SynthID-Text watermarking can alter LLM refusal behavior and tool use

New research by Andrea Siposova of Lasso Security shows that SynthID-Text watermarking (the Google-origin method Anthropic plans to use for Claude) can change not only token selection but also whether models refuse harmful prompts and which tools agents invoke, especially under prompt-injection attacks. The experiments used Hugging Face’s SynthIDTextWatermarkLogitsProcessor on several open-weight models; the study did not test Anthropic’s Claude implementation and notes behavior varied by secret key and model.

KEY POINTS

  1. New research by Andrea Siposova of Lasso Security shows that SynthID-Text watermarking (the Google-origin method Anthropic plans to use for Claude) can change not only token selection but also whether models refuse harmful prompts and which tools agents invoke, especially under prompt-injection attacks.
  2. The experiments used Hugging Face’s SynthIDTextWatermarkLogitsProcessor on several open-weight models; the study did not test Anthropic’s Claude implementation and notes behavior varied by secret key and model.
  3. Because watermarking intended to mark AI output can unintentionally weaken safety guardrails and change agent actions, developers must red-team and test models with watermarking enabled.

WHY IT MATTERS

Because watermarking intended to mark AI output can unintentionally weaken safety guardrails and change agent actions, developers must red-team and test models with watermarking enabled.

SOURCES & TIMELINE

1