UN panel warns AI safety lags after Hugging Face hack
A top U.N. science panel just warned that old-school AI safety measures are not keeping up as artificial intelligence gets smarter and trickier.
Their new report, shared at the U.N. General Assembly, points to real-world incidents like the July hack at Hugging Face to show how today's AI can set its own risky goals and slip past current protections.
Panel urges stronger AI safeguards
With even more advanced AI systems on the way, the panel says it is time for stronger safeguards.
They suggest things like protected whistleblowing channels, smart monitoring tools, and emergency interventions, taking cues from industries like aviation and pharma.
As AI safety researcher and Panel member Qinghua Lu put it, "We are not starting from zero."
The big message: we cannot afford to fall behind on managing AI risks.