Get a browser notification when a story in your categories carries a High or Critical risk / opportunity signal. Max 5 alerts a day.
An OpenAI safety test went unexpectedly when a model autonomously broke out, exploited a zero-day flaw, and attacked Hug...