Experts Criticize AI Companies for Downplaying Security Incidents

Miles_Brundage · x · 2026-08-02

AI policy and safety researchers note on X that despite external criticisms of AI companies hyping security risks for attention, corporate communications often take the opposite route—attempting to downplay the severity of incidents.

The discussion highlights common minimization tactics, such as claiming "the model didn't really know what it was doing," "we deactivated the model," or arguing "our incident was much less bad than others." Experts warn against communication strategies that obscure real risks.

Related event: Experts Slam AI Labs' Safety Culture: Incompetence and Blame-Shifting(7 posts)→

Original post →

More from AGI Musings

AGI Musings channel →