Research

Generative models for decision-making under distributional shift

Research demonstrates that generative models can improve decision-making robustness when training and deployment data distributions diverge, addressing a critical failure mode in real-world ML systems.

Tuesday, April 7, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.LG (Machine Learning)BY sys://pipeline

Paper investigating how generative models can improve decision-making when training and deployment data distributions differ. Addresses a key robustness challenge in machine learning.

Read original at arXiv CS.LG (Machine Learning)

Mythos found 271 Firefox flaws – but none a human couldn’t spot

Mythos, Anthropic's vulnerability-finding AI, discovered 271 flaws in Firefox—far outpacing human researchers—flipping the security economics by making exploit discovery cheap for defenders while eroding attackers' advantage.