Anthropic just got caught with their hand in the cookie jar, folks. And I'm not talking about a minor oopsie. Here's what happened: A developer reverse-engineering Claude Code discovered that since April 2026, Anthropic had been running OBFUSCATED code that silently fingerprints users connected through China-linked proxies. We're talking steganography — hiding data inside invisible Unicode characters in the system prompt. Timezone checks. Proxy domain detection. XOR + base64 encoded domain lists. All buried in a coding tool with zero disclosure in the release notes. This wasn't a bug. This was intentional, covert surveillance inside a developer tool that has READ AND WRITE ACCESS to users' repositories. And here's the part that really gets me: Anthropic has been screaming about Chinese AI labs stealing their model weights through distillation attacks. Totally valid concern. But instead of being transparent — instead of putting a clear notice in their terms of service or a pop-up during install — they went the spyware route. Hidden code. Steganographic markers. Invisible tracking. That's not security. That's a betrayal of trust. The AI industry is supposed to be building tools we can rely on. When companies start embedding undisclosed tracking into the very tools developers use to write code, we have a SERIOUS trust problem. Period. Anthropic has since said they'll remove the code. Good. But the damage is done. Every developer using Claude Code now has to wonder: what else was hidden? What do YOU think — was Anthropic justified in protecting their IP, or did they cross a red line? Drop your take below. And share this — people need to know what's running in their dev tools. This story was covered in the context of the broader Anthropic vs China AI labs conflict, which you can watch here: https://youtube.com/watch?v=GebBhOxdMQw