Users are reporting that Claude is becoming difficult to use for legitimate scientific research, particularly in biology and immunology. The AI's safety guardrails are frequently flagging even basic scientific queries, hindering its utility as a research assistant. This contrasts with its ability to assist with code related to bioweapon development tools like RFdiffusion and AlphaFold, which do not trigger the same flags. AI
IMPACT AI safety guardrails may be overly restrictive, impacting legitimate scientific research and potentially creating blind spots for harmful applications.
RANK_REASON User feedback on a product's usability and safety features.
Read on HN — claude cli stories →
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →