US Government Urges Meta to Submit AI Model for Security Evaluation
The United States government is urging Meta to submit its artificial intelligence (AI) models to undergo security evaluations.
According to reports from Engadges on Wednesday (24/6) local time, Meta is the only major technology company that has not yet voluntarily submitted its AI models for government review. Through this evaluation process, the US government aims to analyse the capabilities of these AI models while identifying any potential vulnerabilities.
Several other AI companies are known to have cooperated with the government previously, such as OpenAI and Anthropic, which have already undergone testing for AI models that have not yet been released to the public. Meanwhile, Google, xAI, and Microsoft have also agreed to grant access to their latest AI models to the Centre for AI Standards and Innovation, an agency established by the Joe Biden administration to evaluate AI technology. This agency is currently led by US Secretary of Commerce Howard Lutnick.
“We share the same goal as the Trump administration to advance US leadership in developing strong and safe cutting-edge AI. We are currently discussing the details of the cooperation and hope to sign the agreement soon,” said Brennan.
Meta launched its latest AI model, Muse Spark, last April. The model features two operating modes: “Instant” and “Thinking”. The Thinking mode allows the system to perform deeper reasoning processes before providing answers to users.
The US government is reported to have sent several official requests to Meta via email. On 2 June, President Donald Trump signed an executive order instructing the federal government to develop an evaluation framework for AI technology before it is released to the public.
Amidst the rapid development of the global AI industry, the US government is said to be increasing its vigilance regarding new technologies. In mid-June, the US government ordered Anthropic to close access to its Mythos 5 and Fable 5 AI models for all foreign nationals on national security grounds. Anthropic subsequently blocked access to both models to ensure compliance with the government directive.
It is noted that Mythos is the company’s most advanced cybersecurity AI model, available only to specific partners, while Fable 5 is designed to bring some of the capabilities of Mythos to a wider user base.
Source: jawawa.id (https://jawawa.id/newsitem/us-government-urges-meta-to-submit-ai-model-for-security-evaluation-1782366938)
Originally published by Antara (Indonesian) (https://www.antaranews.com/berita/5622367/pemerintah-as-desak-meta-serahkan-model-ai-untuk-evaluasi-keamanan), translated from Indonesian