Trump could sign AI executive order as soon as Thursday

Source Domain: www.yahoo.com

Here is a summarized list of key points from the article regarding the White House’s possible voluntary government review of new artificial intelligence models:

– The White House is poised to release an executive order as early as Thursday which will encourage a voluntary review of new AI models by the government before their public release.
– The order includes a voluntary agreement where AI companies would share advanced models with the government before their launch.
– There is a contention over the review period length, with a draft proposing a 90-day pre-launch review, while some companies prefer a shorter review period such as 14 days.
– Key AI companies, including OpenAI and Anthropic, are actively engaged in discussions about the executive order with the White House.
– An early review of advanced AI models could help the government identify and mitigate potential cybersecurity threats.
– The draft executive order includes two main sections: one focusing on cybersecurity and another defining specific “covered frontier models” eligible for the early review.
– The cybersecurity section proposes forming a voluntary “clearinghouse” to detect and fix vulnerabilities, along with increased hiring at the US Tech Force.
– The White House is moving towards more regulatory oversight on AI, different from the hands-off approach taken by the previous Trump administration.
– Anthropic and OpenAI have already begun practices of offering early, controlled access to their AI models to businesses and governments for cyber defense purposes.
– The Department of Commerce had prior initiatives for major tech companies to share AI models with the government for national security evaluations, although this initiative is no longer listed on their official website.