thinker

active

thinker:hubinger-et-al

Hubinger et al.

Cited for demonstrating models can be trained with persistent backdoor deceptive behaviors (Sleeper Agents)

Authored

0

Introduces

0

Studies

1

Affiliations

0

Cited by

0

More papers — OpenAlex / S2

Studies (1)

Backdoor Behavior (Sleeper Agents)

Other inbound relations (1)

citesWhen Thinking LLMs Lie: Unveiling the Strategic Deception in Representations of Reasoning Models(paper)

Recent mentions (1)

papers-typed
wang-2025-thinking-llms.md