Anthropic banned 11.4M Claude accounts in six months; appeals succeed only 10.5% of the time
新智元 · wechat · 2026-10-11
Anthropic's July transparency report reveals 11.4 million Claude accounts were banned in H1 2026 — over 60,000 per day. Of 398,000 appeals filed, only 42,000 succeeded (10.5%). Bans stem from extreme content violations (15,000+ NCMEC reports, new ML classifiers), bulk black-market accounts, and region checks. Developers report Kafkaesque appeals: template rejections with no cited clause, permanently frozen appeal pages, auto-closed tickets, and accounts killed 15 minutes after signup. With agents and coding workflows now centered on Claude, bans can wipe out weeks of accumulated work. New restrictions take effect November 12, 2026, and community repos now track survival tips — but the lesson is: don't bet everything on one AI provider.
More from Models
- Google testing new Flash-Lite checkpoints as Gemini Pro tier stays frozen for 235 days — lyraxana · 2026-10-12
- Users Complain Gemini Web Locked Into Auto Mode With Only Flash-Lite — Slow_Pangolin_5494 · 2026-10-12
- Haiku 5.5 costs 75% less than 4.5, but shines only as a subagent under Opus 5.5 — Scobleizer · 2026-10-12
- Power user burns $6,428 in API-priced tokens on two $200 plans: 'VC-subsidized scooter era' — beffjezos · 2026-10-12
- A 'continuity recipe' for making LLMs keep a persistent identity goes viral — RileyRalmuto · 2026-10-12
- ChatGPT Plus $0 first-month offer keeps declining all cards, likely risk-control block — Aggravating-Pea4806 · 2026-10-12