US Government Urges Meta to Submit AI Models for Safety Evaluation
L'essentiel
- The US government is reportedly pressuring Meta to submit its AI models for safety and security evaluation, making it the only major AI player not to have voluntarily done so.
- Other tech giants like OpenAI, Anthropic, Google, xAI, and Microsoft are already cooperating with the Center for AI Standards and Innovation, established by a recent executive order.
Résumé généré par IA
Pourquoi c'est important
The US government is urging Meta to submit its AI models for evaluation due to growing safety and security concerns about AI technologies, with other major AI players already cooperating. President Trump signed an executive order on June 2 to establish a framework for the federal government to evaluate AI releases.
The US government is reportedly urging Meta to submit its AI models for evaluation amid growing safety and security concerns about the risks posed by the latest artificial intelligence technologies. According to The New York Times, Meta is the only major AI player that hasn't voluntarily turned its models in for review. Specifically, the government wants to assess its capabilities and identify vulnerabilities.
Both OpenAI and Anthropic are already working with the government to test unreleased models, Reuters said. Meanwhile, Google, xAI and Microsoft have already agreed to provide the Center for AI Standards and Innovation with early access to their new models. The agency, created by the Biden administration and staffed with technical experts to vet AI technology, is headed by Commerce Secretary Howard Lutnick.
"We share the [Trump] administration's goal of advancing US leadership on robust and secure frontier AI," said Meta spokesperson Francis Brennan. "While we are working through the details, we hope to sign the agreement soon." The government has been sending its requests to the company via emails, The Times has reported.
President Trump signed an executive order on June 2 to establish a framework for the federal government to evaluate AI releases. Under the order, the government has until the end of July to develop a review process. Still, the goal is for companies to give authorities up to 30 days to evaluate their technologies before they're made public. While there's no official review process yet, the aforementioned companies, except Meta, have reportedly been sharing their models with the government on their own for months now.
Meta launched its latest model, Muse Spark, back in April. It has "Instant" and "Thinking" modes, with the latter enabling reasoning capabilities. With the Thinking mode activated, Muse Spark will take a few extra moments to reason through a prompt for a more thorough answer. While it's not quite as powerful as other companies' frontier models, the government has been keeping a closer eye on the AI industry recently.
À surveiller
Perspective IA — des possibilités, pas des certitudes
Meta will sign the agreement to submit its AI models for evaluation.
Probable · En quelques semaines
Questions ouvertes
- When will Meta sign the agreement?
- What specific vulnerabilities will be identified?
- How will the government's review process function?
