Daily AI Briefing

Tuesday, September 29, 2026

The 6 top stories of the 55 AIBriefs curated that day.

Get this briefing in your inbox every morning

LaunchAI Models15 sources

Meta releases Muse Spark 1.3 agentic coding model

Meta Superintelligence Labs' fourth Muse Spark release in five months uses ~20% fewer tool calls and ~25% fewer tokens than Muse Spark 1.2. Pricing is $0.2 vs $25 per 1M output tokens versus Claude Opus 5, with a 90%+ discount if you opt in to training.

EventPolicy6 sources

OpenAI says Moonshot-linked accounts ran reasoning extraction campaign

OpenAI disrupted a coordinated distillation campaign that spiked to 16,000 extraction attempts from over 4,000 users on July 24-25, 2026, with related prompt-pattern activity across 15,000+ users. OpenAI says operators manipulated model interactions rather than breaking encryption, and banned the accounts.

EventBusiness15 sources

AMD to acquire Fei-Fei Li's World Labs for $8.2B

AMD reached a definitive agreement to buy the spatial-intelligence lab in an all-stock deal valued at about $8.2 billion, expected to close by end of 2026 pending regulatory approval. Fei-Fei Li will join AMD as EVP and chief scientist, reporting to Lisa Su.

EventBusiness12 sources

Google's Project Suncatcher launches first TPU satellite on SpaceX Falcon 9

The refrigerator-sized MVP satellite carries four Google TPUs and about one kilowatt of solar power, built by Planet Labs and launched on the Transporter-18 rideshare. The TPU fires in 15-minute bursts to avoid straining power and thermal systems; Google's target is a network of 81 satellites flying in close formation.

LaunchCybersecurity7 sources

Anthropic launches Cyber Mission with free OSS Scanner

Anthropic's Cyber Mission starts with two efforts: OSS Scanner, free periodic security scans of open-source projects by its strongest models, and the Critical Infrastructure Defense Program, which brings frontier models, on-site engineers, and threat research to operational-technology defenders. Project Glasswing partners have uncovered at least 129,000 verified flaws.

AnalysisRobotics2 sources

GPT-6 Astra controls Unitree G1 humanoid in unseen kitchen

A Stanford project (tml.stanford.edu/homebody/) let GPT-6 Astra drive a Unitree G1 humanoid in a kitchen it had never seen, tidying rooms and fetching objects from vague requests. A separate demo turns a room video into an interactive 3D world for robot training.

Also that day

  1. AnalysisEpoch AI: Cost of a Given AI Performance Level Falls 47% Per Quarter
  2. EventInstinct raises $1B Series C at $10B valuation
  3. LaunchOpenAI launches Astra for Law, a GPT-6 offering for legal work
  4. EventGoogle fights EU push to open Android to rival AI bots
  5. EventMistral opens Munich hub for Physics AI and Industrial AI
  6. LaunchMicrosoft Research introduces Quine, an AI system for biology
  7. EventAnthropic signs $11.6B, seven-year cloud deal with Akamai
  8. LaunchAlibaba Qwen releases Qwen-Audio-3.1-Realtime voice model
  9. LaunchBlack Forest Labs releases FLUX 3 Action, a 7B open-weights robot model
  10. LaunchDeepSeek Harness v0.2 ships official desktop apps
  11. EventMother Jones: ChatGPT safeguards bypassed before Tumbler Ridge shooting
  12. AnalysisOpenAI publishes data on coding agents accelerating its own research

37 more stories that day — open the live feed.

Daily brief

Get tomorrow's AI brief in your inbox

AI News Briefing for Tuesday, September 29, 2026 — AIBriefs