Meta's Muse is quietly generating deeply personal AI training data at scale
signulll · x · 2026-09-16
The author argues Meta is generating enormous amounts of deeply personal, actually actionable AI training data through Muse — what people see, want, ask, choose, buy, ignore, and act on, especially via connectors. He compares it to Onavo, but built in-house. The compounding data flywheel is only beginning, making Muse effectively "Facebook 2.0" as a company — which is why Zuckerberg had to go all-in.
More from Companies & People
- Runway CEO: Usage of video and image models is exploding — 'the time is now' — c_valenzuelab · 2026-09-16
- AI lets firms build compounding engines of institutional knowledge, says new essay — chetanp · 2026-09-16
- Alexandr Wang: Muse Was Nat Friedman's Passion Project — alexandr_wang · 2026-09-16
- Anthropic Engineer Explains FDE: How Palantir Hits $4M Avg Contract Value — dotey · 2026-09-16
- AI company employees, including top researchers, are pressuring CEOs on AI risk — hlntnr · 2026-09-16
- Desiree resigns from Anthropic and sounds the alarm bell — AIandDesign · 2026-09-16