An open model from China beat Claude on a security test -- at a sixth of the cost
Semgrep ran GLM 5.2 against Claude on a narrow vulnerability-finding task and the free, open-weight model came out ahead for far less money.
OpenAI showed off GPT-5.6 -- then handed the guest list to the US government
Three new models, strong enough at hacking that OpenAI is only letting about twenty vetted partners in, at the government's request.
The ban on Anthropic's most powerful model just partially lifted -- for Americans only
About a hundred U.S. institutions regain access to Mythos and Fable, while foreign nationals stay locked out and rival labs rush to fill the gap.
Put AI agents in charge of a Civilization game and they reach for the nukes
A new benchmark let language-model agents play Civilization VI -- and they learned that the fastest path to winning ran straight through mutually assured destruction.
Smart glasses fed students live exam answers -- and schools have no idea how to stop it
A cheating ring at Brown used AI-connected glasses to pipe real-time answers onto the lenses, and universities across Asia are seeing the same thing.
The world's central-bank watchdog warns an AI bust could spill into the wider economy
The BIS cautions that if the AI funding boom unwinds, the damage may not stay contained to tech -- it could ripple into growth and credit.
A model that rivals the frontier now squeezes onto a single high-end desktop
Aggressive compression shrinks GLM 5.2 by more than 80 percent while keeping most of its accuracy, putting a near-frontier model within reach of local hardware.
A security writeup catalogs how AI agents get attacked -- and one claim raised eyebrows
A semi-annual review tallies fresh ways to attack AI agents, from prompt injection to token leakage -- alongside one extraordinary, unverified extraction claim.