The Daily AI Briefing

Tuesday 13 October 2026

AI news

Meta introduces Muse Wonder and Muse Wonder Full

Meta released Muse Wonder through its Model API and Muse Code on 12 October; Muse Wonder Full is available to Muse Code subscribers. The models add Active Memory across conversations and Active Focus for sustained tasks. Consumer Muse rollout is still forthcoming, and Meta says a fuller safety report will follow.

Why it matters: This new model release advances beyond the earlier context-model and agent-protocol announcements, while developer access arrives before the broader consumer rollout.

about.fb.com

Microsoft puts Maia 200 into production and previews Maia 300

Microsoft says Maia 200 now serves production workloads including Copilot, with GitHub Copilot and Foundry planned to follow by year-end. It plans multi-gigawatt Maia deployment and Maia 300 production from 2027. The company projects 40% more tokens per megawatt for Maia 300 than leading GPU systems and up to seven times Maia 200’s rack throughput.

Why it matters: The new production and hardware roadmap is separate from yesterday’s decision-model launch, making power efficiency a concrete constraint on expanding agent services.

blogs.microsoft.com

Google expands AI activity tracking from 10 to over 100 activities

Google’s October 12 update expands automatic activity detection from 10 to more than 100 activities, with rollout later this month. It reports nearly twice the accuracy in offline comparisons with its previous model. Health Guardian trend reports also arrive later this month for eligible Premium subscribers; these features are for wellness and are not intended for medical purposes.

Why it matters: This new wearable rollout is distinct from Google’s earlier work-agent releases, showing how contextual AI is moving into recurring personal tasks with specific access limits.

blog.google

ngram 2.0 launches with explainers, screencasts and AI presenters

ngram 2.0 launched on Product Hunt on 12 October, combining animated explainers, screen-recording walkthroughs and AI avatar presenters. Its agent takes documents or recordings through scripting, narration and visual editing. Makers say changed product interfaces currently require an updated recording; automatic GitHub-driven updates are still planned.

Why it matters: Bringing three video workflows into one release gives teams a practical way to explain complex material, while retaining a manual update step when the underlying product changes.

producthunt.com

Nonprofit AI

Middle Powers AI grant call

Astralis Foundation and Macroscopic Ventures have an open AI policy funding call with around $10 million available and typical grants of $100,000 to $2 million. Nonprofits, researchers, universities and informal teams can apply for work helping countries such as Australia shape a safer AI transition. Applications close on 24 October at 5:59 p.m. AEDT and are assessed on a rolling basis.

Why it matters: This gives civil society and research teams funding for policy and institution-building, rather than requiring them to develop a frontier model.

middlepowers.ai

PAI mental health partners

Partnership on AI launched an AI and Psychological Well-Being Working Group on 8 October. Mental-health and crisis-support organisations will work with researchers, AI practitioners and people with lived experience on chatbot responses to suicide or self-harm risk. Draft guidance is planned for public input later this year, followed by final guidance organisations can adopt.

Why it matters: This new working group is a separate development from PAI’s previously covered funding, giving frontline services a role in shaping guidance before it is finalised.

partnershiponai.org

Charity workforce AI white paper

Zoe Amar Digital, Corndel and Microsoft published a charity-workforce white paper on 5 October. It combines the 2026 Charity Digital Skills Report with a June leadership roundtable to examine roles, entry-level jobs and skills through 2036. The landing page describes a practical checklist for leaders, trustees and funders; access requires a details form.

Why it matters: A workforce plan can connect AI adoption to recruitment, training and core funding, without treating reused survey figures as new measurements.

corndel.com

Localgiving AI toolkit webinar

Localgiving’s 1 October announcement offers a 60-minute Zoom demonstration on 21 October for charities and community interest companies. CEO Andrei Yusfin will show its AI Toolkit taking an idea through project planning, funding discovery and grant drafting. Registration is open; the announcement reports no measured fundraising gains.

Why it matters: The session offers small organisations a concrete workflow to evaluate, while keeping a product demonstration separate from evidence of improved grant outcomes.

localgiving.org

One moment

Loading the Hub…