Specifically, the paper estimates that Llama 3.1 70B has memorized 42 percent of the first Harry Potter book well enough to reproduce 50-token excerpts at least half the time. (I’ll unpack how this ...

Meta's Llama model has memorized Harry Potter and the Sorcerer's Stone so well that it can reproduce verbatim excerpts from ...

Meta’s Llama 3.1 is repeating chunks of copyrighted books like Harry Potter, raising serious legal concerns over AI training ...

Meta is paying $14.3 billion to buy 49-percent of Scale AI and hire its CEO, Alexandr Wang, to reboot its troubled AI efforts ...

Once poised to rival GPT-4.5 and Claude 3, Meta’s most powerful LLM is now delayed, highlighting the steep challenges of building next-gen AI.

Meta is making an investment in artificial intelligence company Scale and recruiting Chief Executive Alexandr Wang to join a ...

Meta on Saturday released the first models from its latest open-source artificial intelligence software Llama 4, as the company scrambles to lead the race to invest in generative AI. But the ...

Meta is forming a superintelligence team to help build its AI efforts. The recruitment is overseen by Zuckerberg.

However, controversy erupted almost immediately following Saturday's debut ... as the prior Llama release, 3.1, showed a lot of leadership by the Meta team in advancing AI techniques to make ...