Why Some Courts Are Banning AI Tools in Legal Cases – And What It Means for Your Privacy

Why Some Courts Are Banning AI Tools in Legal Cases – And What It Means for Your Privacy

If you use a public AI chatbot like ChatGPT for work or personal tasks, you may want to reconsider what you feed it. Several U.S. judges have recently issued orders prohibiting the use of public generative AI tools during the discovery phase of litigation. The reason? Data submitted to these tools can be absorbed into training sets, potentially exposing sensitive information.

These rulings aren’t just a niche concern for lawyers. They highlight a broader privacy risk that applies to anyone handling confidential data—financial details, medical history, trade secrets, or even personal conversations. Here’s what happened, why it matters, and how you can protect your own information.

What happened

In recent months, at least three federal judges have entered orders that bar parties in civil cases from using publicly available AI tools like ChatGPT, Google Gemini, or Microsoft Copilot (the consumer version) when processing or reviewing discovery materials. Discovery is the pretrial stage where both sides exchange evidence, documents, and other sensitive records.

The concern, as reported by Bloomberg Law News, is that any data entered into these tools may be used to train future versions of the model—or may be stored and reviewed by the provider. One judge specifically noted that information submitted to a public chatbot “does not remain private,” even if the user believes their session is confidential.

These orders do not apply to enterprise or private instances of AI tools—versions that contractually guarantee your data won’t be used for training. The distinction is critical.

Why it matters for everyone, not just lawyers

The principle behind these bans extends well beyond courtrooms. Any time you paste a confidential email into ChatGPT to summarize it, upload a sensitive spreadsheet, or ask a free bot to analyze private health data, you may be handing that information over to be stored or used to improve the model.

This is not a hypothetical risk. OpenAI, for example, has stated that ChatGPT’s free and Plus tiers can use conversations for training unless you opt out (and even then, certain metadata may still be retained). Similar policies apply to many free-tier models from Google, Anthropic, and other providers.

For individuals, the risk includes:

• Exposing personal identifiable information (PII) that could be linked back to you.
• Revealing trade secrets or business strategies.
• Sharing medical or financial data without adequate legal protection.

For professionals handling sensitive information—journalists, doctors, accountants, HR managers—the stakes are even higher. A single copy-paste into a public model could violate confidentiality agreements, professional ethics, or data protection laws like GDPR or HIPAA.

What readers can do

You don’t need to stop using AI tools. But you should adopt a privacy-first approach.

1. Know your tool’s data policy. Before entering anything sensitive, check whether the service stores or trains on your data. Look for clear language in the privacy policy or terms of service. If it says “your interactions may be used to improve our models,” assume your input is not private.

2. Use enterprise or private versions. Many providers offer paid tiers with data-use guarantees. For example, ChatGPT Team, Enterprise, and API usage (when you enable privacy controls) typically promise not to train on your data. Google Workspace’s AI features for business accounts also have stricter privacy protections.

3. Anonymize or synthesize data. When you need help drafting or analyzing something, remove names, account numbers, addresses, and any other identifying details before inputting it. For brainstorming, use placeholders instead of real facts.

4. Consider local or offline models. Tools like Llama 3, Mistral, or GPT4All can run on your own computer. Your data never leaves your device. This is the most private option for sensitive work.

5. If you must use a public tool for legal or professional work, get written permission. Check with your compliance officer, legal counsel, or IT department first. Don’t assume it’s allowed.

Sources

This article draws from the Bloomberg Law News report “Judges’ Public AI Bans During Discovery Zero in on Privacy Risk” (June 2026). That report identified multiple court orders restricting public AI use in discovery and explained the data-use rationale. For current model privacy policies, also refer to OpenAI, Google, and Anthropic’s official documentation, which may change over time.

Final note: AI privacy is a moving target. What’s true today may change tomorrow as terms of service and regulations evolve. Always verify the current policy of any tool before trusting it with sensitive data.