OpenAI has upgraded the potential biological risk level for its latest artificial intelligence (AI) model, implementing additional safeguards as a “precautionary approach.”
The AI firm on Thursday released ChatGPT agent, a new agentic AI model that can now perform tasks for users “from start to finish,” according to a company press release.
OpenAI opted to treat the new model as having a high biological and chemical capability level in its preparedness framework, which evaluates for “capabilities that create new risks of severe harm.”
“While we don’t have definitive evidence that the model could meaningfully help a novice create severe biological harm—our threshold for High capability—we are exercising caution and implementing the needed safeguards now,” OpenAI wrote.
“As a result, this model has our most comprehensive safety stack to date with enhanced safeguards for biology: comprehensive threat modeling, dual-use refusal training, always-on classifiers and reasoning monitors, and clear enforcement pipelines,” it added.
OpenAI’s newest model, which began rolling out to various paid users last week, comes as tech companies increasingly turn toward the agentic AI space.
Perplexity released an AI browser with agentic capabilities earlier this month, while Amazon Web Services (AWS) announced new tools last week to help its client build AI agents.
The ChatGPT maker’s latest release comes as the company plans to open its first office in Washington to boost its policy ambitions and show off its products, according to Semafor.
