Fable/Mythos safeguards are overly strict
Users are reporting that Anthropic's new Fable and Mythos models have overly strict safety guardrails, hindering legitimate research. A biologist noted that prompts related to entire fields of biology, even non-dangerous ones like genome assembly, are being rejected. This user expressed concern that these stringent safeguards will disproportionately affect less-funded research areas, potentially limiting scientific progress. AI
IMPACT Overly strict safety measures in AI models could hinder scientific research and innovation in critical fields.