Abundant: Claude safety guardrails flagged RSI tasks but zero refusals actually blocked agents

Madisonkanna · x · 2026-11-20

Abundant's @quasicoh shared observations from running Claude on the FrontierRSI benchmark: despite Anthropic having published that it restricted RSI-type (self-improvement) capabilities on some models, not a single safety refusal actually prevented an agent from completing a task.

The observation, surfaced via @MTSlive's repost of a question from @codyavant about how fallbacks and refusals factor into scores, suggests current RSI safety restrictions are fairly brittle against autonomous agent retries.

Original post →

More from Models

Models channel →