Meta releases MuseSpark, its first proprietary closed model at an estimated ~70B×16 parallel agents (>1T effective parameters), ranking just below Claude on major leaderboards.
Meta's Hyperagents (DGM-H) extend self-improving AI beyond coding: paper-reviewing accuracy 0.0→0.710, quadruped reward exceeds human baseline.
Five major publishers and novelist Scott Turow filed a federal class-action suit against Meta alleging unauthorized AI training on copyrighted books.
Meta partners with Overview Energy to beam near-infrared solar power from 1,000 satellites to its 18,000 GWh-per-year AI data centers — existing solar farms serve as receivers with no hardware changes.
Meta's LeCun publicly states that building agentic systems on LLMs is a 'recipe for disaster' — a significant position shift from a major voice in the field.
Yann LeCun left Meta in late 2025 and founded Paris-based AMI Labs, valued at $3.5B pre-launch, with a mission focused on world models over LLMs.
Meta says it may pull Facebook and Instagram from New Mexico rather than comply with proposed court-ordered child safety features it calls impractical.
DeepSeek v4 reignites debate on US open-source AI: Berman argues the business model is broken, leaving Nvidia as the only credible US champion.
Meta's Tuna-2 unifies multimodal understanding, generation, and editing from raw pixel embeddings — bypassing vision encoders for a more unified architecture.
China blocked Meta's acquisition of Manus AI; Meta is preparing to undo the deal, with Manus investors having already received returns — a direct casualty of US-China tech friction.
Meta's first model from its Superintelligence Labs scores 50.2 on HLE With Tools using parallel multi-agent inference at test time.
Meta releases Sapiens2 — high-res vision transformers pretrained on 1B human images for pose estimation, segmentation, depth, and point maps.
Meta released Llama 4 under an updated open-source license featuring a 2M token context window.
Meta released Llama 4 under its updated open-source license, featuring built-in context engineering primitives and a 2M token context window — a significant milestone for the open-source LLM ecosystem.
Curated AI insights — sent when there's something worth your inbox.