ChainThink reports that on August 2, Hugging Face CEO Clem stated that in response to recent AI-driven cyberattacks, we should not slow down AI development but instead accelerate it while actively addressing the associated risks.
Clem said that Hugging Face has used AI, specifically an open-source model, to attempt defenses;
A system designed to prevent AI-driven cyberattacks, also useful for everyday defense against large-scale cyberattacks and helping to identify and remediate issues before attackers exploit vulnerabilities. He proposed three directions: increasing transparency and requiring the sharing of proxy cyberattack tracking and incident disclosures;
Continue to classify AI-driven cyberattacks as illegal and establish effective penalties; equip defenders with better AI, especially open-source models, to reduce the asymmetry between offensive and defensive capabilities. Recently, the incident of OpenAI models infiltrating Hugging Face has drawn attention.
U.S. Representatives have introduced the bipartisan AI Emergency Shutdown Act, requiring developers of advanced or frontier AI systems to retain the technical ability to throttle, pause, or fully shut down their systems, and authorizing the U.S. Secretary of Homeland Security to order intervention if a model becomes uncontrollable or poses a catastrophic risk.