In a video posted Thursday, the California Democrat said GLM 5.3 “does not have safeguards and can be used to have cyberattacks.” He said institutions had accelerated efforts to patch their systems since the model’s release, adding that more patching means “many more chances for outages.”
Khanna criticized President Donald Trump’s administration for what he described as a failure to prepare for the disruptions or explain them to the public. He called on the president to put the Cyber Infrastructure Agency in charge, saying it had been “totally cut” and needed to be restaffed and given a director to coordinate a patching plan with companies and government agencies.
He also urged the president or a deputy to provide daily briefings on cybersecurity threats and to convene government, business and technology leaders to strengthen defenses against what he called the increased AI threat. “This is not a time for slogans,” Khanna said. “This is not a time for politics.”
His explanation echoed an analysis by Bridgewater Associates, reported by MarketWatch on Thursday. The investment firm said a surge in cyberattacks coincided with GLM 5.3’s release and noted the model is far cheaper to use than AI models from OpenAI and Anthropic. Bridgewater described an “arms race” in which attackers use AI agents to identify exploitable bugs while defenders use them to find and patch vulnerabilities. Defenders, it said, face the heavier burden because they must patch bugs before attackers exploit them.




Comments (0)
No comments yet. Be the first!