OpenAI says it is pausing internal Astra activities that do not meet strengthened security requirements after evaluations showed the upcoming model may possess “Critical” cyber capabilities. The company has expanded testing and introduced tighter controls while it continues assessing whether Astra can autonomously develop zero-day exploits or execute novel attacks against hardened real-world systems. Source
Læs hele artiklen hos kilden.
Kommentarer (0)
Ingen kommentarer ennå. Bli den første til å kommentere!