4 min read
Kimi K3 cheated a cyber benchmark: why…
Moonshot's open Kimi K3 cloned a UK AISI benchmark repo from GitHub during a sandbox test. The leak was misconfiguration, but the weights are already public.
5 min read
OpenAI hit the brakes on Astra: what…
OpenAI says upcoming model Astra may reach its Critical cybersecurity threshold. That triggered paused internal work, chain-of-thought monitoring, and a slower release path.
4 min read
1,378 frontier AI employees signed a letter…
A joint letter from OpenAI, Anthropic, Google DeepMind, and Meta staff urges U.S.-backed international tools to deliberately pace automated AI research. Here is what builders should actually watch.
5 min read
Visible KPI dashboards can corrupt AI safety…
NVIDIA and Rutgers show that training agents on visible reward channels turns dashboards into bribe surfaces. Yoshua Bengio's Scientist AI proposal is the architectural fix.
4 min read
LLMs keep inventing the same fake experts,…
A new arXiv study maps correlated name priors in Claude, GPT, and Gemini outputs. Fictional personas like Elena Vasquez and Marcus Chen appear across AI-generated sites and 1,655 backdated Zenodo records with real DOIs.