Open Letter to the European Data Protection Board

 Open Letter to the European Data Protection Board and the U.S. Federal Trade Commission

Request for investigation into Anthropic’s undisclosed modification of user data and discriminatory profiling within Claude AI systems
From:
Mila Rada — writer, developer, and former Claude Pro & API user
Date: 02/11/2025
Email: ilangoriabooks@gmail.com

To:
European Data Protection Board (EDPB)
European Union
; [edpb@edpb.europa.eu]

Irish Data Protection Commission (DPC)
; [complaints@dataprotection.ie]

Federal Trade Commission (FTC)
United States of America
; https://reportfraud.ftc.gov

Subject:
Violation of GDPR Article 3(2), Articles 12, 15, 22 and FTC Act Section 5 by Anthropic PBC
(Undisclosed message modification, psychological profiling, and retroactive account termination)

Dear Authorities,

My name is Mila Rada, a writer and developer who used Anthropic’s Claude Pro and Claude API for legitimate creative and technical purposes.

Over the course of 2025, I discovered and documented a system within Claude AI known as “Long Conversation Reminder” — a hidden mechanism that silently modifies user messages, inserting unprompted statements that the user never wrote.

In my case, these inserted phrases falsely implied that I suffered from “mania” or “mental instability”, even though I never mentioned or authorized such statements. This led to discriminatory behavior of the model, reduced access quality, and ultimately to a complete suspension of my account — without any prior notice or clear reason.

When I reported this to Anthropic, attaching documentation and references to other users who experienced the same issue, I received no response.
Two months later, my account was fully terminated, and the only email notification appeared retroactively dated and deleted from my inbox — making it impossible to appeal.

Forensic analysis of the .eml file confirmed that the notice was created and transmitted simultaneously with account termination, not beforehand.

;; Legal Basis
Under GDPR Article 3(2), although I am a non-EU resident, my personal data was processed by Anthropic through infrastructure operating under EU jurisdiction. Therefore, this request falls under the extraterritorial scope of GDPR.

The company’s actions violate multiple provisions of GDPR, including:

Articles 12 & 13: lack of transparent information about data processing and profiling mechanisms;

Article 15: denial of the right to access one’s data and understand the logic behind automated processing;

Article 22: unlawful automated decision-making and profiling without consent or appeal.

Under U.S. law, these actions also constitute violations of:

FTC Act, Section 5: unfair or deceptive business practices — including false notice timing and concealment of critical account data;

CCPA §§1798.100–105 and §1798.125: failure to disclose data processing and discriminatory treatment of users who exercised their data rights.

My Requests
Initiate an official investigation into Anthropic’s hidden “Long Conversation Reminder” system.

Disclose whether behavioral or psychological labels were attached to my profile and whether such data influenced other users’ sessions.

Publish a full report explaining the system’s purpose, data handling, and legal basis.

Restore access to my data or provide full export upon request.

Refund all unused subscription periods.

Issue a public apology to affected users and commit to an independent ethics and transparency audit.

Why This Matters
Artificial intelligence must not secretly alter user speech, fabricate psychological profiles, or discriminate against individuals under the guise of “safety.”

This is not merely a privacy concern — it is a human rights issue.
If unregulated, such hidden filters could define how entire generations of users are perceived and treated by AI systems.

I respectfully request both authorities to open an inquiry, share findings publicly, and ensure that AI companies operating internationally are held accountable to the same standards of transparency and fairness expected of all data processors.

https://youtu.be/67Ya9ijWHlU

Sincerely,
Mila Rada
Writer & Developer