Redwood Research

QuestionHow well can agents perform useful privileged work, sabotage it, and detect sabotage in a common environment?

Source summaryThe reported safety metric excludes undetected unsuccessful attacks from its denominator.

NUWA research note · independent review pending · Contains: question / method / findings / interpretation / limits / connections

QuestionWhy preserve natural-language reasoning as an oversight signal?

Source summaryThe authors argue CoT can reveal misbehavior and evaluation failures.

NUWA research note · independent review pending · Contains: question / method / findings / interpretation / limits

QuestionCan a multi-step agent be kept useful while stopping a deliberately subversive policy under a limited audit budget?

Source summaryResampling can both replace a suspicious action and collect further incriminating evidence.

NUWA research note · independent review pending · Contains: question / method / findings / interpretation / limits / connections

QuestionWhat evidence could justify that a monitored internal agent will not exfiltrate sensitive information even if it tries?

Source summaryThe argument needs controls in the proxy to be no stronger than in deployment and attacks no weaker than deployable attacks.

NUWA research note · independent review pending

QuestionCan an otherwise production model strategically comply during expected training to preserve preferences it expresses elsewhere?

Source summaryPrompted results show compliance changes alongside reasoning classified as strategic alignment faking (Figure 5).

NUWA research note · independent review pending · Contains: question / method / interpretation / limits

QuestionHow do agents compare with experts on open-ended ML research engineering?

Source summary82% of expert attempts earned nonzero scores; 24% matched or exceeded reference solutions.

NUWA research note · independent review pending · Contains: question / method / limits

QuestionIf deceptive conditional behavior is already present, do standard safety-training methods reliably remove it?

Source summarySome backdoors persist after these interventions.

NUWA research note · independent review pending

QuestionCan a defender obtain useful work from an untrusted stronger model while limiting harm from an adaptive adversarial policy under a restricted audit budget?

Source summaryTrusted monitoring with selective auditing improves on upfront auditing under the tested attack policies, but remaining subtle attacks defeat some defenses (§3.2).

NUWA research note · independent review pending · Contains: question / method / findings / interpretation / limits