Anthropic Claude Code Leak Shows 512k Lines & Pet
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Anthropic released Claude Code 2.1.88, but a packaging error included a source map file exposing over 512,000 lines of its TypeScript codebase, as users reported on X and Reddit.
- Users uncovered a Tamagotchi‑style pet that sits beside the input box and reacts to coding, and a “KAIROS” feature that could enable an always‑on background agent.
- Anthropic later fixed the issue, stating no sensitive customer data was exposed and attributing the leak to human error, not a security breach.
- GitHub repository containing the leaked code has amassed more than 50,000 forks after users copied it.
- Arun Chandrasekaran of Gartner warned the leak could help bad actors bypass guardrails but sees limited long‑term impact, calling it a prompt for Anthropic to improve operational maturity.
Why it matters: Anthropic must tighten its release processes to prevent future leaks, while developers gain unprecedented access to its proprietary code, and security analysts warn that the exposed internals could be weaponized to sidestep Claude Code’s safety guardrails, this could erode user trust and force Anthropic to allocate resources to remediation and compliance.
