Meta releases AI research models

The company released key elements of five AI models that are part of its Chameleon suite, under a research-only licence. These include image-to-text and text-to-music generation models, a multi-token prediction model and a technique for detecting AI-generated speech.

“As the field rapidly innovates, we believe that collaboration with the global AI community is more important than ever,” a company statement said.

“By publicly sharing this research, we hope to inspire iterations and ultimately help advance AI in a responsible way.”

The move came days after Stefano Fratta, global engagement director on Meta privacy policy, announced that training of Meta’s large language models (LLMs) using public data had been put on hold in the UK and Europe.

The pause was the result of a request from the Irish Data Protection Commission (DPC).

Fratta said: “We are committed to bringing Meta AI, along with the models that power it, to more people around the world, including in Europe. But, put simply, without including local information we’d only be able to offer people a second-rate experience. This means we aren’t able to launch Meta AI in Europe at the moment.”

Meta is also dealing with concerns from the UK regulator, the Information Commissioner’s Office (ICO).

Meta has said its models identify patterns rather than identify specific people or their information, and that the company does not train its systems on private messages between friends and family members, not on any data generated by under-18s.