Meta to grant EU institutions access to AI model Llama for national security purposes

Tech giant Meta said it will make its artificial intelligence (AI) tool Llama available to “key democratic US allies” in Europe and Asia, amid a push to develop advanced AI for national security reasons.

Meta Platforms, which owns Facebook, Instagram, and WhatsApp, said it will extend access to the open-source AI models to France, Germany, Italy, Japan, South Korea, NATO, and the European Union.

It had previously made Llama available to US defence and national security agencies and private companies in the sector, as well as those in Australia, Canada, New Zealand, and the United Kingdom.

“The US and its closest allies should have the best tools at their disposal to defend themselves and keep their citizens safe,” Meta said in a blog post Tuesday.

Llama models have helped US agencies with “decision-making, mission-specific capabilities, and operational efficiency,” the company said, adding that it could offer Llama to additional countries in the future.

Notably, Meta’s terms of service bar Llama from being used for “military, warfare, nuclear industries or applications, [and] espionage”.

However, when it launched Llama in the United States last year, Meta said it was in the “wider democratic world’s interest” for American AI models to prevail over those developed in China.

That announcement came just days after Reuters reported that researchers in China had reworked an older version of Llama to support the country’s own military.