SignalFrontier releaseSG-0029
Meta released Muse Wonder and Muse Wonder Full, models it says lead on personal and professional agent tasks
Meta made Muse Wonder available through its Meta Model API and Muse Code, and a stronger Muse Wonder Full to Muse Code subscribers only. Meta's AI chief said the model was trained to keep trying other approaches when blocked rather than stall. Meta's model card says it cannot rule out the unmitigated model reaching its high-risk threshold in the chemical and biological domain, so it added safeguards. It rates cyber and loss-of-control risk moderate or lower even before safeguards.
Why it matters
Meta is bringing a model trained to push past obstacles to its Muse agent app, which can handle messages and private documents, and where users already say the agent read messages without permission, which Meta disputes. In Meta's own hacking test, 71 of the 92 flags the model captured came through flaws the testers had not planted. The risk ratings are Meta's own, and its full safety report is not out yet.
Sources
Read the reporting
3 sources. Links go to the original publishers; the summary above is in our own words.
- newsMeta debuts Muse Wonder AI model in latest test of users' willingness to payCNBC · Oct. 12, 2026cnbc.com/2026/10/12/meta-muse-wonder-ai-model-agent.html
Signals
More from the news desk
Cite and share
Use this signal
Citation
Paperclip Index. “Meta released Muse Wonder and Muse Wonder Full, models it says lead on personal and professional agent tasks.” Signal SG-0029. Reported Oct. 12, 2026; updated Oct. 12, 2026. https://paperclipindex.com/signal/SG-0029