Google DeepMind Warns AI Cyberterror Could Leak Your Smart Toaster’s Secrets
KEY POINTS
- •Demis Hassabis, Google DeepMind CEO, spoke at Axios’ AI+ Summit in San Francisco on October 24, 2025.
- •Hassabis warned that cyberterrorism targeting energy or water infrastructure is 'probably almost already happening.'
- •He called Google’s cybersecurity focus vital, noting his personal estimate of AI catastrophe risk ('p(doom)') is 'non-zero' and needs serious mitigation.
At Axios' AI+ Summit in San Francisco this Thursday, Google DeepMind’s Demis Hassabis casually dropped that AI-powered energy or water cyberterror (yes, cyberterror) is 'probably almost already happening'—which sounds way less comforting than your neighborhood neighborhood BBQ. His May prediction that Artificial General Intelligence (AGI) might birth by 2030-level scary, now comes with a 100% less sci-fi, 100% more 'please don’t hack my home' vibe. Hassabis, CEO of the AI brain trust, highlighted Google’s cyber-defense obsession, admitting his own p(doom)—the probability of AI catastrophe from misuse—is 'non-zero.' That’s tech-speak for 'we might be toast, but we’re also making toast smart.'
Share the Story
Source: Axios | Published: 12/5/2025 | Author: Ben Berkowitz