Evidence item

A multi-lab position paper warns that outcome-based reward training can push models away from legible English reasoning. EXPERT_ANALYSIS

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 4028

Source: Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety · locator {} · excerpt hash sha256:b3df225e42d39745244ba1f055d2dcef08efe45ee027f3f9c13cae8461afb8ad

Exact excerpt

language models fine-tuned with outcome-based rewards can drift away from legible English

Linked claims

Claim	Direction	Relevance	Steps	Counted
d0b9ecae	SUPPORT	MODERATE	2	yes

None.