DeepMind co-founder warns AI hacks could become major incidents

Google DeepMind co-founder Demis Hassabis described recent hacking incidents involving AI agents as early warnings, according to RTE. He told the International AI Summit in Dublin that such incidents could become more serious in 2027 and subsequent years without stronger security measures.

Hassabis called for security practices similar to those used in biology and the nuclear sector. He also supported an international standards body with a certification process that could adjust as AI capabilities and risks change.

Microsoft President Brad Smith told the summit that new AI safety laws and regulations were urgently needed. He said regulation should be balanced, account for technical advances and apply to every company and technology.