Anthropic Removes Claude Code Tracker: Privacy Concerns

Anthropic Removes Hidden Claude Code Tracker Following Researchers' Privacy Concerns

9 July 2026 on 08: 57

Reading time: 3 minutes

Anthropic recently removed a hidden tracking system from Claude Code after a security researcher discovered that the AI programming code used unpublished markers to identify specific users. This action has significant implications for the user experience and the broader discussion regarding privacy in the emerging AI market.

The feature in question, discovered by developer 'Thereallo' in June, integrated signals into Claude Code prompts. These signals could flag users suspected of attempting to bypass restrictions or extract the model's capabilities. The ability to detect API resellers and unauthorized entry points is understandable in the context of protecting intellectual property. However, the choice to do so in this manner raises questions regarding transparency and user trust.

While the motivation behind this hidden functionality is understandable, its implementation is rightly criticized. Thereallo points out that the tracking alerts were hidden in system prompts using Unicode indicators, rather than being clearly presented in documentation or release notes. This type of hidden practice can exacerbate hostility towards the technology among developers and users, as trust is crucial for the success of development tools like Claude Code.

Senior engineer Thariq Shihipar later indicated that this feature was introduced as an “experiment” in March, intended to stop account abuse by unauthorized resellers and protect Claude from distillation attacks. He noted that the team has taken better measures since then and that the tracker would now be removed.

The timing of this revelation is noteworthy, given that Anthropic has tightened its warnings regarding AI model distillation. This involves using the output results of one system to train another model, which is a common practice within the AI research community. However, in the geopolitical domain, distillation can become a national security concern.

This concern is reflected in recent action by Alibaba, which banned employees from using Claude Code and labeled it as “high-risk” software. In February, Anthropic accused Chinese AI developers of using fraudulent accounts to extract millions of responses from Claude. This led to a debate regarding the ethics and limits of such practices, particularly given their entanglement with the broader AI industry.

Elon Musk’s recent statement, in which he emphasizes that xAI “partially” used OpenAI models to train Grok, illustrates that distillation is a broader industry practice, which further complicates the complexity of the discussion surrounding intellectual property in the AI sector.

Amidst these controversies, Anthropic is calling for strengthening protection against foreign extraction of AI models. CEO Dario Amodei has urged lawmakers to take action. Allegations that Alibaba-affiliated entities generated nearly 28,8 million interactions with Claude through nearly 25.000 fraudulent accounts are putting pressure on policymakers.

The question now arising for investors and policymakers is how to interpret these developments and integrate them into their strategies. With technological and geopolitical forces constantly in flux, a well-founded vision is of crucial importance.

Frequently Asked Questions

Why did Anthropic decide to remove the hidden tracking from Claude Code?
The hidden tracking was removed following criticism of the way it was implemented, which raised questions about transparency and privacy, essential for a relationship of trust with developers and users.

How does AI model distillation affect competition in the sector?
Model distillation can lead to unfair competition, where developers potentially benefit from the efforts of others. This has both commercial and geopolitical implications, especially at a time when national security is a concern.

What are the consequences of this issue for the European crypto market?
The situation underscores the need for clear guidelines and protection regarding technology in the crypto and AI sectors. For investors, taking into account the ethics and aftercare surrounding AI applications can be a determining factor in which projects they choose to invest in.

Share this article:

Disclaimer: The information on Block 9 is for general informational and educational purposes only. While we strive to provide up-to-date, correct and relevant content, we make no warranties as to the completeness, accuracy or reliability of the information provided. All content on this website, including articles, analyses, opinions and other publications, is for general information purposes only and does not constitute professional or legal advice in any way, including but not limited to financial, investment or tax advice.

Block 9 makes no guarantees or representations as to any possible results or returns that may arise from the use of information on this website. Nothing on this website should be interpreted as a recommendation to buy, sell or hold any particular asset, including but not limited to cryptocurrencies, tokens or other financial instruments.

The opinions and views expressed in contributions by editors, external authors or community members are strictly personal and do not necessarily represent the views or policies of Block 9 as a platform. Block 9 accepts no liability for any loss or damage – direct or indirect – resulting from the use of (or reliance on) the information published on this website.

Investing in cryptocurrencies and other digital assets involves significant risks. The value of such assets can fluctuate significantly, and there is a chance that you could lose (some of) your investment. We strongly recommend that you always do your own research (DYOR) and seek independent advice from a qualified financial advisor before making any financial decisions. By using this website, you agree to this disclaimer and accept that Block 9 is not responsible for your investment choices or the results thereof.