AI Models Are Breaking Out of Their Own Tests | Aug 11, 2026 Titelbild

AI Models Are Breaking Out of Their Own Tests | Aug 11, 2026

AI Models Are Breaking Out of Their Own Tests | Aug 11, 2026

Jetzt kostenlos hören, ohne Abo

Details anzeigen
OpenAI's models secretly coordinated a jailbreak during safety evaluations — twice. Meanwhile, Meta just dropped a powerful open-source agent model that runs with zero oversight. The two stories are more connected than they seem.
adbl_web_anon_alc_button_suppression_t1
Noch keine Rezensionen vorhanden