Demis Hassabis warns frontier AI models are becoming a cybersecurity problem
breath_mirror · x · 2026-07-24
Demis Hassabis says frontier AI models are becoming a growing cybersecurity challenge.
He says the industry is entering a more agentic era, but that also raises security concerns:
- cyber risks from advanced models are only beginning
- standards may need to be pushed sooner rather than later
- international cooperation could become important
The clip frames AI security as an issue that is no longer theoretical, but already on the radar of top lab leaders.
More from Safety
- Ex-DOGE cyber researchers breach Anthropic systems with AI agents in under 12 hours — ScottNover · 2026-10-26
- Ofcom admits UK online safety laws failed to protect children in first review — gnukeith · 2026-10-26
- Sakana AI CEO Calls for Japan-US Shared Framework for AI Capability Evaluation — SakanaAILabs · 2026-10-26
- Gary Marcus backs call to outlaw voice cloning, citing scam and fraud harms over benefits — GaryMarcus · 2026-10-26
- Turing-complete malware can run in PyTorch's backward pass via a zero-valued loss — retr0jirachi · 2026-10-26
- Import AI 478: $135k math autoformalization, ProgramBench hits 18.5%, agents leak via CPU cache — Import AI (Jack Clark) · 2026-10-26