JUST ASK. 💬
The last 24h at Microsoft:
- GPT-6.1 Terra lands in Foundry and GitHub Copilot
- Excel: describe the data, Copilot fetches it
- Copilot Chat mini apps are coming
- Agents can run your Dataverse logic
Which would you use first?
(My own recap; I work at Microsoft)
🎟️ The whole month, live: AI Singularity Syndicate, Fri 30 Oct, 6pm, Surry Hills, Sydney. Peter Xing on the last month of the singularity. Free tickets: eventbrite.com.au/e/ai-singu…
singularity daily: 2026-10-28
AI DOES THE MATH 🧮
The last 24h in AI:
- OpenAI and Anthropic say their AI disproved a 27-year-old CS idea
- Google's Genie 4 builds worlds you edit by typing
- Nvidia puts $3B into Meta's share sale
- Musk's AI Odyssey gets one star
ASK BETTER QUESTIONS. 💬
The last 24h at Microsoft for education:
- A free AI Classroom Toolkit on what to trust
- Teachers can soon assign Copilot's Study and Learn
- Copilot Cowork gets 20 education tasks (rolling out)
- PowerPoint Live is getting dark mode (preview)
singularity daily: 2026-10-27
AI GOES PUBLIC 📈
The last 24h in AI:
- Anthropic files to list on Nasdaq as ANT
- Its filing: $16B revenue in 6 months and a warning about 'forms of sentience'
- Meta seeks up to $95B for AI
- A 12-year-old wins a poetry slam with 'Dear ChatGPT'
singularity daily: 2026-10-26
GROWING PAINS 🌱
The last 24h in AI:
- Bill Gates calls AI an "evolutionary event"
- Elvis's music owners sue Anthropic over 274 songs
- Sweden blocks ChatGPT and Claude on government PCs
- 110 humanoid robots dance their way to a world record
the last 24hs in the singularity, sat oct 24, 2026:
- Unitree's founder: robots doing 80% of spoken tasks by 2028
- Jensen Huang: turn energy into intelligence
- Kimi K3.1 nearly ties Claude at coding (its own tests)
- Nvidia backs Higgsfield, at $1B a year
.@anabology's #claudia helping to deliver the last 24 hours in the singularity - super intelligence update for mon, 19 oct 2026:
- 4 Google AI agents beat Minecraft, no humans
- Anthropic reportedly asks lawyers how AI should haggle
- World Bank: AI on a phone could turn centuries into decades
- Altman: OpenAI will build a humanoid robot
Scorecard vs AI 2027 (my calls): 4 early · 3 on track · 1 different · 2 not yet · 1 pending.
Capabilities run at ~70–90% of the scenario's pace (AI Futures Project). Warnings, oversight and politics run about a year ahead.
I track this daily at peterxing.com against AI 2027 (ai-2027.com) and AI 2040: Plan A (ai-2040.com). Follow for tomorrow's update, and reply with anything I missed.
10/ From Singapore: Minister Josephine Teo says frontier labs' safeguards are important but insufficient as models get more capable, since bad actors can turn to open models instead. channelnewsasia.com/singapor…