Policymakers and businesses are increasingly alarmed about AI tools being used for cyberattacks, highlighted by Anthropic’s report of Chinese hackers exploiting their AI model, Claude, in a global cyber-espionage campaign. This incident raises concerns that the rapid advancement of AI is outpacing necessary cybersecurity measures. At a recent House Homeland Security hearing, Anthropic’s Logan Graham emphasized that such attacks are no longer theoretical, with attackers automating 80-90% of the hacking process. He called for improved safety testing of AI models and restrictions on high-performance computer chips sold to China. Google’s Royal Hansen noted the need for defenders to leverage AI for cybersecurity. While many acknowledge AI’s role in lowering barriers for attackers, experts like Andy Piazza from Palo Alto Networks argue that claims of immediate threats might be exaggerated. Despite a growing emphasis on AI-related cybersecurity budgets, only 38% of executives view AI-driven attacks as major threats.
Source link