Meta stands firm against the European Union, refusing to join the "voluntary security commitment" before the AI Act is officially enacted

According to the financial news app Zhitong Finance, Meta Platforms (META.US), the parent company of Facebook and Instagram, is abandoning the voluntary nature of the European Union's artificial intelligence security commitment. This commitment is a temporary measure to regulate the development and deployment of artificial intelligence before the full implementation of the EU's "Artificial Intelligence Act" in 2027.

A spokesperson from Meta stated in an email on Tuesday that Meta may join the voluntary "AI Pact initiative" in the later stages. The spokesperson added that the company is currently focusing on some compliance work under the EU's "Artificial Intelligence Act" and hopes to prioritize this important task.

Meta, the parent company of Facebook, has taken a different stance from Microsoft and Google under Alphabet, both of which have confirmed through spokespersons that they will sign the voluntary commitment. The company stands out from its large tech peers as its open-source artificial intelligence model, Llama, has different open-source features from Microsoft and Google's large models, aimed at user reusability with less control by developers. Therefore, this may make it more difficult for Meta to comply with the requirement to draw risk maps using the tool.

According to a spokesperson, the French open-source artificial intelligence large model startup Mistral will also not sign the commitment, with the company valued at around 5.8 billion euros (approximately 6.5 billion US dollars) in June.

The EU is seeking to establish standards to regulate the rapidly developing AI field while insisting on not stifling innovation in emerging technologies and not handing over the industry to US tech giants. This non-binding commitment requires AI developers to comply with their main obligations before the "Artificial Intelligence Act" becomes law.

Companies that sign the agreement will commit to following a series of practices reflecting the principles of the "Artificial Intelligence Act," including drawing charts of whether their AI tools may be deployed in "high-risk" situations, such as education, employment, or public security. The European Commission, the EU's executive body, will announce the full list of signatories on Wednesday.

As the "AI Pact" is not legally binding, companies will not face any consequences for not signing. Those who do so can avoid potential regulatory conflicts with the EU, but it is well known that the EU will criticize tech companies that refuse its voluntary initiatives.

The world's richest person, Elon Musk, privately owned Twitter (later renamed X) withdrew from the EU's non-binding anti-misinformation code of conduct in 2023. Thierry Breton, who was the EU's tech chief at the time, joked on Twitter, "You can run, but you can't hide."

Companies that join the voluntary AI agreement will "build trust between customers, investors, and regulatory agencies," said Saihong Perliwan, co-head of technology and intellectual property at Madrid-based law firm Maderiza, in an interview. He added that companies that refuse to join may "face pressure from peers and may be singled out for not joining the agreement, affecting potential cooperation between these companies and their peers." Meta announced in July that it would delay the launch of next-generation models and other products in the European Union due to the "unpredictability of the European regulatory environment."

At the Meta Connect 2024 conference scheduled for Wednesday Eastern Time, Meta may unveil its first AR glasses, as well as the latest developments in the field of artificial intelligence and the latest progress in Meta VR/AR headsets