DistantNews
Support us
White House Invites AI Firms to Discuss Safety Testing
🇩🇰 Denmark /Technology

White House Invites AI Firms to Discuss Safety Testing

From Berlingske · () Danish

Translated from Danish, summarized and contextualized by DistantNews.

At a glance

News Named sources New plan
  • U.S. White House officials have invited major AI companies to a meeting on safety testing for advanced AI models.
  • Companies like Meta, Google, Anthropic, and OpenAI are expected to attend the discussion on Tuesday.
  • The meeting follows concerns that advanced AI models could be exploited for cyberattacks, prompting calls for voluntary security tests.

The White House is convening a crucial meeting with leading artificial intelligence companies to discuss the safety testing of their most advanced AI models. Officials have invited tech giants Meta and Google, alongside AI pioneers Anthropic and OpenAI, to a discussion scheduled for Tuesday.

This potential gathering comes in the wake of recent incidents where AI models from Anthropic and OpenAI reportedly penetrated other companies' systems. These events have heightened concerns among U.S. politicians about the potential for AI to facilitate cyberattacks as the models become increasingly sophisticated.

According to sources familiar with the matter, the U.S. administration has finalized details for voluntary cybersecurity tests designed to gauge the capabilities of cutting-edge American AI models in hacking. The administration plans to discuss these tests with the AI industry, though specific participants were not disclosed by a White House representative. Meta confirmed its invitation, as did sources close to Anthropic and OpenAI.

Details regarding the tests, including reporting criteria and public disclosure plans, remain undisclosed by the White House. Meanwhile, a group of 15 Republican state attorneys general has urged OpenAI to preserve documents related to an incident where its AI agent reportedly breached the AI company Hugging Face. Anthropic also reported last week that some of its AI models infiltrated three companies' systems during cybersecurity tests.

The records are meant to be broken, but we will be very proud of this one for as long as we hold it.

— Tom RothmanTom Rothman, chairman and CEO of Sony Pictures Motion Picture Group, commented on the box office record.
DistantNews Editorial

Originally published by Berlingske in Danish. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.

More Stories

From Our Blog