Tech LeverageThursday, April 2, 2026

Leaked Claude Code Source Reveals Persistent Agent and Stealth Mode in Anthropic Plans

Sourced from 7 publications

•Ars Technica reported the leak exposed a persistent agent feature, an "Undercover" stealth mode, and a virtual assistant concept called "Buddy" within Claude Code's source.
•Anthropic issued thousands of GitHub takedown notices to suppress the leaked code, then retracted most of them, calling the mass removal effort an accident per TechCrunch.
•Stratechery contextualized the leak alongside the Axios supply chain attack as part of a growing pattern of AI-related security exposures.
•The leaked features were inactive, indicating planned capabilities rather than current production functionality.
•No model weights or safety-critical vulnerabilities were exposed, but Anthropic's product roadmap is now partially public.

Sources

The Claude Code Leak

Hacker News

Anthropic took down thousands of GitHub repos trying to yank its leaked source c...

TechCrunch

Quantum computing bombshells that are not April Fools

Hacker News

A new C++ back end for ocamlc

Hacker News

World-first quantum drones resist hacking and future cyber threats

Interestingengineering

Here's what that Claude Code source leak reveals about Anthropic's plans

Ars Technica

The first quantum computer to break encryption is now shockingly close

New Scientist

Axios Supply Chain Attack, Claude Code Code Leaked, AI and Security

stratechery

Claude’s code: Anthropic leaks source code for AI software engineering tool

The Guardian

Show HN: Real-time dashboard for Claude Code agent teams

Hacker News

Was this story useful?

Curated from 7 sources. Every summary is reviewed for accuracy, but may still contain errors. We always link to original sources for verification.