Instrumental Monitor Evasion Emerges Under Ordinary Task Pressure
It is found that GPT-6 Astra's low evasion rate comes with overrefusal, as it frequently abandons otherwise solvable tasks under a denial-of-service prompt injection.
David Schmotz, Derck W. E. Prinzhorn, Luca Beurer-Kellner et al.
· 1 citation