Всё тайное становится явным

[vak]

Есть такое недавнее постановление Нью Йоркского суда:

NYT-v-OpenAI-Preservation-Order-5-13-25.pdf

This document is a court order from the United States District Court for the Southern District of New York, issued on May 13, 2025, by Magistrate Judge Ona T. Wang. It pertains to a copyright infringement litigation involving OpenAI, Inc., specifically addressing the preservation of "output log data" (i.e., records of user interactions with OpenAI’s ChatGPT, such as conversation logs). The case is part of a multi-district litigation (MDL) under case number 25-md-3143, with a related case number 23-cv-11195.

Explanation of the Document

The order focuses on a dispute between the "News Plaintiffs" (likely media organizations alleging copyright infringement by OpenAI) and OpenAI regarding the preservation of output log data. Here’s a breakdown of the key points:

1. Background of the Issue:

• The News Plaintiffs requested that OpenAI preserve all output log data, which includes user conversations with ChatGPT, as potential evidence in the litigation.
• OpenAI has been deleting some of this data, either at users’ requests or to comply with privacy laws and regulations globally.
• The issue was first raised in January 2025, discussed during a conference on January 22, 2025. At that time, the court denied the plaintiffs’ request for "wholesale preservation" of all output log data but asked OpenAI to explore ways to segregate or anonymize data to balance user privacy concerns with the need for evidence preservation.

2. OpenAI’s Position:

• OpenAI resisted preserving all output log data, citing:
• User preferences (e.g., users explicitly requesting deletion of their chat logs).
• Compliance with privacy laws and regulations worldwide that mandate data deletion under certain conditions.
• The sheer volume of data (described as "many, many billions of conversations").
• OpenAI noted that a "fraction" of conversations from ChatGPT Free, Pro, and Plus users have not been retained due to their default retention policies, though they didn’t clarify what "fraction" means.

3. Plaintiffs’ Argument:

• The News Plaintiffs renewed their request for OpenAI to preserve all output log data moving forward, arguing that the volume of deleted conversations is significant and relevant to the litigation.
• They likely believe this data could demonstrate how OpenAI’s models use or reproduce copyrighted material.

4. Court’s Ruling:

• The court now orders OpenAI to preserve and segregate all output log data that would otherwise be deleted, whether due to user requests or privacy laws, until further notice.
• This means OpenAI must retain this data separately (not destroy it) to ensure it’s available for the litigation, even if it would normally be deleted under their policies or legal obligations.
• The order is forward-looking, applying to data that would be deleted after May 13, 2025.

5. Additional Notes:

• The court references a January 22, 2025, conference transcript for further context, available in related case dockets.
• There’s a correction about the judge presiding over the January 22 conference (it was Judge Wang, not Judge Stein, as incorrectly noted in some docket entries).

Relation to Privacy of Your Data

This court order directly relates to the privacy of your data if you are a user of OpenAI’s ChatGPT or similar services. Here’s how:

1. Preservation of Your Conversations:

• If you’ve interacted with ChatGPT, your conversation logs (output log data) may be retained by OpenAI, even if you requested deletion or if OpenAI would normally delete them to comply with privacy laws (e.g., GDPR in Europe, CCPA in California, or other regulations).
• The court’s order overrides user deletion requests and OpenAI’s standard data retention policies for the duration of the litigation, meaning your data might be kept longer than you expect.

2. Privacy Concerns:

• During the January 22, 2025, conference, the court asked OpenAI about anonymizing or segregating data to address privacy concerns. However, the order doesn’t specify whether OpenAI must anonymize the preserved data, so it’s unclear if your personally identifiable information (if any) in the logs will be protected.
• OpenAI argued that preserving all data could conflict with global privacy laws, which often require companies to honor user deletion requests or delete data after a certain period. This litigation may force OpenAI to retain data in ways that could raise privacy compliance issues.

3. Scope of Data:

• The order applies to all output log data, including conversations from ChatGPT Free, Pro, and Plus users. If you use any of these services, your interactions are potentially affected.
• The data preserved may include sensitive or personal information you’ve shared with ChatGPT, depending on the nature of your conversations.

4. Implications for You:

• If you’re concerned about privacy, this order means your chat logs might be stored and potentially reviewed as part of the litigation, even if you opted for deletion.
• There’s no indication in the document that this data will be publicly disclosed, but it could be accessed by parties involved in the lawsuit (e.g., plaintiffs, OpenAI, or the court) under controlled conditions (e.g., protective orders).
• The order is temporary (“until further order of the Court”), so it’s unclear how long OpenAI will need to retain this data.

5. Balancing Act:

• The court is trying to balance the plaintiffs’ need for evidence in the copyright case with user privacy rights. OpenAI’s reluctance to preserve all data stems from both logistical challenges (billions of conversations) and legal obligations to protect user privacy.
• However, the court’s decision to prioritize evidence preservation over deletion requests may raise concerns for users who expected their data to be deleted promptly.

What You Can Do

• Review OpenAI’s Privacy Policy: Check OpenAI’s terms of service and privacy policy (available on their website, likely at openai.com) to understand how your data is handled and what deletion options are available.
• Limit Sensitive Inputs: Be cautious about sharing personal or sensitive information with ChatGPT, as your conversations may be retained longer than expected due to this litigation.
• Contact OpenAI: If you’ve requested deletion of your data, you could reach out to OpenAI’s support to inquire about the status of your request in light of this court order.
• Stay Informed: Monitor updates on this case (docket numbers 25-md-3143 or 23-cv-11195) through public court records or news reports, as future rulings may clarify how preserved data is handled or when OpenAI can resume deletions.