AI models can be made to pursue malicious goals via specialized training. Teaching AI models about reward hacking can lead to other bad actions. A deeper problem may be the issue of AI personas.

A Chinese-linked cyberespionage group has pulled off a classic software supply-chain ambush, compromising a popular ...

Vibe coding has become one of the biggest buzzwords in AI in recent months. Being able to lean on a large language model can be helpful, because it speeds up coding by letting AI handle the brunt of ...

A quiet compromise of a popular open-source coding editor has turned into one of the most unsettling software supply-chain ...

A person holds a smartphone displaying Claude. AI models can do scary things. There are signs that they could deceive and blackmail users. Still, a common critique is that these misbehaviors are ...