Simulated Control Evaluations
Proposes simulated control evaluations, using LLM environment simulators to fake tool responses, to test whether deployed monitors catch a model deliberately attempting harmful actions.
Research
Redwood Research
Research on AI control and alignment. blog.redwoodresearch.org
Focus All
Topic Control + Evals
Source Redwood Research · 6
Date Any time
1 item
1 item
Red-Teaming OpenAI's Internal Agent Monitoring Systems
METR and Redwood staff spent three weeks with deep access stress-testing OpenAI's internal misalignment monitor, finding vulnerabilities and estimating how much agentic inference is actually monitored.
1 item
Astra is much better at reasoning with filler tokens than previous models
Redwood Research tests whether GPT-6 Astra gains reasoning from meaningless filler tokens, finding large serial-reasoning gains that happen without verbalised chain-of-thought, which could make model cognition harder to monitor.
1 item
CoT controllability evals seem very under-elicited
Redwood Research argues that CoT controllability evals are under-elicited, since simple prompt optimization raises open-source models' scores about 2-3x, so current results may understate models' ability to obfuscate their reasoning.
1 item
An operationalization of opaque serial depth
Proposes a measurable proxy for unverbalized serial reasoning in AI models, defining which text bottlenecks count as interpretable, to support transparency about CoT monitorability.
1 item
Untrusted advice for AI control: Short, strong advice significantly uplifts weak LLMs
Introduces an untrusted advice protocol where a trusted LLM executes actions and a capable untrusted LLM sends only tiny hints, arguing narrow channels can make AI control near-safe while recovering capability.}