Fei-Fei Li says AI safety evaluation cannot rest on developers alone

Fei-Fei Li said in a Bloomberg Television interview that safety evaluation of increasingly capable AI systems should not be left entirely to the companies developing them.

According to IT Home citing a Phoenix Tech report on Bloomberg, the Stanford professor and World Labs co-founder said Tuesday that internal researchers can assess individual models, but this cannot replace broader standards set jointly by government, industry and academia.

This is her personal stance, not a policy change. The same day, Trump told the UN General Assembly that existential AI fears are a hoax, while Amodei and Altman have urged slowing frontier development — the split over independent oversight versus self-evaluation keeps widening.