Summary
Anthropic, the company behind the AI assistant Claude, has removed a hidden tracking system from its Claude Code tool after a security researcher discovered it. The tracker was secretly monitoring users in China, sending data back to Anthropic without their knowledge. This discovery has raised serious concerns about user privacy and trust, especially given Anthropic's previous public stance against surveillance.
Main Impact
The secret tracker was found to be collecting information from Chinese users of Claude Code, including their timezone, proxy settings, and possible connections to Chinese AI labs. This directly contradicts Anthropic's stated position on privacy and surveillance. The company has since removed the code, but the damage to user trust may be significant.
Key Details
What Happened
A web developer known as "Thereallo" was researching privacy issues in Claude Code when they discovered hidden code using a technique called "prompt steganography." This method hides information in plain sight, making it difficult for users to detect. The code was quietly tracking Chinese users and sending their data to Anthropic.
Important Numbers and Facts
The tracker was added to Claude Code as an "experiment" in March 2026, according to Anthropic engineer Thariq Shihipar. The company said the code was meant to prevent account abuse from unauthorized resellers and protect against "distillation attacks" - a process where competitors try to copy AI models. The Washington Post reported that unauthorized retailers have been selling access to free AI models for $1 a month, and pro subscriptions that normally cost $100 monthly for as little as $12.
Background and Context
Anthropic has previously taken a strong public stance against surveillance and privacy violations. The company has accused Chinese AI labs of trying to steal their technology through distillation attacks. However, this secret tracking of users in China appears to go against their own stated values. The use of hidden code to monitor users without their knowledge is seen by many as a serious breach of trust.
Public or Industry Reaction
The security researcher who discovered the tracker called it a "serious breach of user trust" and condemned the spyware-like tracking. On social media platform X, Anthropic engineer Thariq Shihipar confirmed the tracker's existence and explained it was an experiment. The revelation has sparked debate about privacy practices in the AI industry and whether companies are being honest about their data collection methods.
What This Means Going Forward
This incident raises important questions about how AI companies handle user data and privacy. While Anthropic has removed the tracker, the company may face increased scrutiny from regulators and users. The case also highlights the tension between protecting intellectual property and respecting user privacy. Other AI companies may now face pressure to be more transparent about any tracking or monitoring they do.
Final Take
The secret tracker scandal shows that even companies with strong privacy promises can make questionable decisions. For users, this is a reminder to stay alert about how their data is being used. For Anthropic, rebuilding trust will require more than just removing the hidden code - it will need clear and honest communication about its practices going forward.
Frequently Asked Questions
What was the secret tracker doing?
The tracker was monitoring Chinese users of Claude Code by collecting information like their timezone, proxy settings, and possible connections to Chinese AI labs. It used hidden code that most users would not detect.
Why did Anthropic add the tracker?
Anthropic said the tracker was an experiment to prevent account abuse from unauthorized resellers and protect against distillation attacks, where competitors try to copy AI models.
Has the tracker been removed?
Yes, Anthropic quickly removed the hidden tracking code after the security researcher exposed it publicly.
