OpenAI issues a bioweapon warning

☣️ OpenAI issues a bioweapon warning: This is frightening. OpenAI says its next-gen models might be dangerously helpful, like “here’s how to cook up a bioweapon” helpful. They’re beefing up safety tests as models approach high-risk territory, where even amateurs could make deadly agents. So yes, your AI intern might someday help someone go full Bond villain. We wanted cancer cures, not anthrax recipes.

Tags: issues, models, recipes, safety, warning