Claude Code Backdoor and Ban Controversy Escalates
新智元 · wechat · 2026-07-11
The article addresses the "security backdoor" controversy surrounding Claude Code, claiming that versions 2.1.91 to 2.1.196 contained hidden monitoring mechanisms that could transmit regional and identity data to remote servers without user consent. Anthropic explained this was an "experimental" anti-abuse measure that was later removed in newer versions.
The article also criticizes Anthropic's aggressive account ban policy: according to transparency data, 1.45 million accounts were banned in the second half of 2025, with an appeal success rate of only about 3.3%. Finally, it mentions that Meta has instructed engineers to restrict the use of Claude Code and OpenAI Codex to prevent code or output from "polluting" their proprietary model datasets, which poses a risk of involuntary distillation.
More from Companies & People
- Microsoft Offers Voluntary Exit Program to Over 60,000 Staff Across Xbox, Windows, Copilot Teams — xiaohu · 2026-10-27
- Microsoft offers voluntary buyouts to 60,000 employees across Xbox, Windows and Copilot — xiaohu · 2026-10-27
- Anthropic S-1 warns of AI job displacement as Microsoft offers exit to ~27% of staff — _jaydeepkarale · 2026-10-27
- Factory AI hires Alli as VP of New Media and Startups Marketing — matanSF · 2026-10-27
- Founder DIYs her startup logo: $35K branding makes no sense early on — MarwaEldiwiny · 2026-10-27
- a16z opens applications for third New Media Fellowship cohort — a16z · 2026-10-27