Analyst argues Meta could actually exclude user data from training, unlike OpenAI's lawyerly wording
ivan_bezdomny · x · 2026-09-16
Analyst ivanbezdomny weighs how AI vendors handle user data for training:
- He argues Meta could genuinely exclude agentic user data from training — it has other data sources and ran a massive end-to-end encrypted messaging app (WhatsApp), so it has real experience with this problem
- Still, he cautions Meta "could also not do it"; it's competence, not trust
- He criticizes OpenAI's wording: posting repeatedly that it's "not technically" training on your data, while leaving room for it to leak into models anyway
The thread highlights the gap between legal fine print and actual data practices across AI labs.
More from Models
- Users accuse OpenAI of silently degrading models for "suspicious" accounts — LiquidVolatility · 2026-09-16
- Rumored mid-tier model Terra looks dead: work splits between Sol/Astra and cheap Luna — kevinkern · 2026-09-16
- IFM's 7B K2-Horizon nearly matches 27B models, tops GPT-5 on BrowseComp, fully open-sourced — karminski3 · 2026-09-16
- Salesforce launches Koa, its first CRM reasoning model, built on NVIDIA Nemotron 3 Super — NVIDIA Blog · 2026-09-16
- Stealth Startup Unveils Non-Chat Model Claiming 100x Speed and Cost Edge Over LLMs — hardimanjames · 2026-09-16
- DeepSeek V4.1 Flash Keeps Timing Out on 2-Hour Agentic Benchmarks, Author Shares Failure Logs — sebnadeau · 2026-09-16