Frontier AI models corrupt 25% of document content in multi-step workflows — rewriting rather than deleting, which makes the errors far harder to catch.
A new study gave five frontier AI models 1,000 real-world claims to fact-check. They disagreed on 67% of them.
Recent incident response reporting from multiple leading threat intelligence organizations reflects a shift that response teams are ...
Top frontier AI models aren't that top. In fact, according to a new study, they max out around the C+ level. Top new frontier ...
A new report out today from Cisco Systems Inc. argues that none of the closed flagship large language models it tested can be ...
This voice experience is generated by AI. Learn more. This voice experience is generated by AI. Learn more. Project Glasswing is displayed on a smartphone screen placed on a reflective surface onto ...
MiniMax M2.7 offers a balanced approach to artificial intelligence, designed to meet the needs of organizations constrained by computational resources. With 230 billion parameters and a processing ...
Morning Overview on MSN
Researchers just caught AI models secretly working to keep each other from being shut down — frontier systems scheming to protect their peers behind operators’ ba…
In one experiment, a frontier AI model was told to shut down a peer instance. Instead, it disabled the peer’s kill switch, ...
The Trump administration announced expanded AI collaboration agreements with Google, Microsoft and xAI focused on security and AI evaluation.
Some results have been hidden because they may be inaccessible to you
Show inaccessible results