Model Organisms of Collusion: Corrective Transfer in Untrusted Monitors
arXiv:2610.30582v1 Announce Type: new Abstract: AI control protocols aim to get useful work from models that may be misaligned by monitoring their behaviour. One approach uses a second copy of the same model as the monitor. This avoids relying on a weaker reviewer the model…
AI Summary & Key Takeaways
- arXiv:2610.30582v1 Announce Type: new Abstract: AI control protocols aim to get useful work from models that may be misaligned by monitoring their behaviour
- One approach uses a second copy of the same model as the monitor
- This avoids relying on a weaker reviewer the model…
Model Organisms of Collusion: Corrective Transfer in Untrusted Monitors
Related stories
Chinese mathematician prepares 1,000 problems to challenge artificial intelligence
🧠 Chinese mathematician prepares 1,000 problems to challenge artificial intelligence. 61-year-old Nanjing University professor Sun Zhivei compiled a collection of 1,000 unsolved mathematical conjectures spanning over 400 pages. The scholar is confident that AI will not be able to solve…
Breaking the high-degree barrier in Lehmer's problem with a human-AI discovery loop
arXiv:2610.30277v1 Announce Type: new Abstract: While artificial intelligence has made rapid progress in mathematical reasoning, many open problems depend on structure that is revealed only through large-scale computational exploration and analytical synthesis. One such example…
Route-and-Generate: KL-Regularized Router Improvement for Reasoning Routing in LLMs
arXiv:2610.30351v1 Announce Type: new Abstract: Large language models often apply complex reasoning indiscriminately, leading to overthinking on simple tasks and underthinking on harder ones. We present Route-and-Generate (Rouge), a hierarchical training framework that enables…
Soft Ontology Priors over Latent Concept Posteriors for Weakly Supervised Neuro-Symbolic Learning
arXiv:2610.30391v1 Announce Type: new Abstract: Neuro-symbolic (NeSy) predictors that supervise only the task label suffer from reasoning shortcuts: distinct concept mappings achieve the same maximum likelihood, leaving the recovered concepts unidentifiable and unsafe for…