OpenAI Pauses Some Work on New AI Model Over Cybersecurity Concerns

The decision reflects internal findings that the upcoming ‘Astra’ model may possess ‘critical cyber capabilities’ and follows a string of AI-testing incidents.