Meta AI / Meta Superintelligence Labs
Meta's AI division, historically the leading open-weights frontier lab (Llama), now pursuing 'personal superintelligence'.
Meta championed open-weight AI with Llama but refused Europe's voluntary code and is now moving toward closed models in its race for superintelligence. Independent graders rank its safety practices near the bottom.
Focus
- Open source
- Safety evaluations
Funding, as far as we know
Corporate (Meta Platforms) - https://investor.atmeta.com/
What they've actually done
Won summary judgment that training Llama on books was fair use on the record presented (Kadrey v. Meta), with the judge warning the ruling was fact-specific
25 June 2025
Refused to sign the EU GPAI Code of Practice, calling it overreach
18 July 2025
Signalled a shift to releasing closed frontier models first under Alexandr Wang's Superintelligence Labs, departing from Llama's open policy
6 April 2026
Graded D (1.10/5) in FLI's Winter 2025 AI Safety Index
2025-12