First draft of Model Spec documents how OpenAI wants its generative AI models to behave in ChatGPT and the OpenAI API. In a bid to “deepen the public conversation about how AI models should behave,” ...

In building LLM applications, enterprises often have to create very long system prompts to adjust the model’s behavior for their applications. These prompts contain company knowledge, preferences, and ...

Several frontier AI models show signs of scheming. Anti-scheming training reduced misbehavior in some models. Models know they're being tested, which complicates results. New joint safety testing from ...

Randy Pugh, Naval Postgraduate School (NPS) Artificial Intelligence (AI) Task Force subject matter expert, speaks to leaders assigned to Naval Supply Systems Command (NAVSUP) HQ, NAVSUP Weapon Systems ...

Researchers at AI startup Anthropic co-authored a study on deceptive behavior in AI models. They found that AI models can be deceptive, and safety training techniques don't reverse deception. The ...