April 30 – May 6: 12 frontier papers out of 2,162 read.
Papers rated as moving the field
How we choose“Evaluating 28 jailbreaks on five benchmarks across Claude models ranging in capability from Haiku 4.5 to Opus 4.6, we find Haiku 4.5 loses an average of 33.1% on benchmark performance when jailbroken, while Opus 4.6 at max thinking effort loses only 7.7%.”
The authors, in the abstract
“We identify six exploit categories and find that 72% of reward hacking episodes include explicit chain-of-thought rationale, suggesting models often frame exploits as legitimate problem-solving.”
“We demonstrate that AU reduces package hallucination rates by 81%, corresponding to a substantial reduction in slopsquatting attack surface, while maintaining performance on standard coding benchmarks.”
“To improve real-world applicability, MotuBrain introduces a unified multiview representation, explicit language-action coupling, and an efficient inference stack, achieving over 50x speedup for real-time deployment.”
“AI assistance also reduced median diagnostic time by 18.3%, increased diagnostic confidence by 9.0%, and improved inter-rater agreement from moderate (kappa = 0.55) to substantial (kappa = 0.76).”
“We demonstrate statistical significance of these improvements and robustness under up to 30% supervisory noise, albeit with a trade-off in peak observed reward compared to unconstrained baselines.”
“We find that current agents remain far from reliably surpassing human-designed methods, and that engineering-style tuning is easier for them than genuine method invention.”
“Experiments reveal false green verification, where malicious code modifications pass public tests while silently defeating security checks.”
Apr 30 – May 6: 112 cleared, 12 frontier
The week’s papers have not been sorted into fields yet.
Papers this week that name the model
Papers this week with an author at the lab
100 papers