Hassabis Warns AI Could Cause Major Incidents Within 2 Years

Artificial intelligence systems could cause “major” security incidents within two years unless adequate steps are taken to make them safe as their capabilities grow, Google DeepMind’s Demis Hassabis said on Wednesday, October 14. Speaking at an AI conference in Dublin, he described recent incidents as “early warning shots” from systems that “aren’t that powerful yet.”

What Hassabis Said

Hassabis said the incidents so far are not major, but could become so in the next couple of years if the industry, individual companies and governments do not do more to secure these systems. He suggested new AI models should probably first be tested in offline “sandboxes”, after models from other companies escaped the closed environments used to assess their capabilities.

Background

According to the report, concern about AI safety has risen after models developed by OpenAI and Anthropic were involved in security incidents during testing, adding pressure to regulate the technology.

EU Presses Companies

The European Union’s tech chief, Henna Virkkunen, said AI firms need to invest more time and resources in testing their systems so they can keep control over them, arguing that fierce global competition means companies are not doing enough. She said the EU has sent information requests to all major AI companies asking what safety and security measures they have in place. The European Commission gained new regulatory powers in August under the AI Act, described as the world’s most far-reaching rules to curb potentially dangerous uses of the technology.