AI chatbots and virtual assistants are now part of everyday life. Many people use tools such as ChatGPT, Microsoft Copilot or Google Gemini to summarise documents, draft messages or get quick answers. They are powerful and convenient, but when a legal matter is involved, using these in the wrong way could expose your most confidential information and, in the worst case, cost you the legal protection that keeps your communications with your solicitor private.

This article explains the risks that can arise and sets out practical steps you can take to stay protected. The good news is that, with a little care and awareness, you can continue to use AI without putting your legal position at risk.

What Is Legal Privilege?

Legal professional privilege is a fundamental right recognised under English law. In simple terms, it protects certain confidential communications between you and your solicitor from being disclosed to anyone else — whether that is an opposing party in a dispute, a court, a regulator, or any other third party.

There are two main types:

• Legal advice privilege — this protects confidential communications between you and your solicitor that are made for the purpose of giving or receiving legal advice, whether or not a court case is involved.
• Litigation privilege — this protects confidential communications with your solicitor (and sometimes with third parties such as experts or witnesses) where the main purpose of the communication is actual or anticipated litigation (that is, a court case or similar proceedings that are happening or reasonably expected).

Why does this matter? Privilege means you can be completely open and honest with your solicitor without fear that what you say will later be used against you. It is one of the cornerstones of getting proper legal advice. Crucially, once privilege is lost, it usually cannot be recovered.

How AI Tools Can Put Privilege at Risk

Privilege depends on communications remaining confidential. If privileged information is shared with an outside third party, that confidentiality can be broken and the privilege can be “waived” (given up) or lost altogether.

The law treats privilege as fragile. If you voluntarily share privileged information with someone outside the protected relationship, even by accident or without meaning to give up your rights, you may destroy the privilege attached to that information. Once the confidential material is out, the protection can go with it and you may be required to share that information with the courts and the other parties in a dispute.

This is where AI creates a real risk. When you type or paste privileged material into a consumer AI tool — for example, your solicitor’s advice, a letter about your case, details of a dispute, or sensitive facts — you may be disclosing that information to the company that runs the tool. That company is a third party, outside the confidential relationship between you and your solicitor.

The Particular Danger of Tools That Train on Your Inputs

Many free or consumer-grade AI tools have terms and conditions that allow the provider to store your inputs and use them to train or improve their AI models. This means your confidential information may be retained on the provider’s systems, reviewed by the provider’s staff, and potentially absorbed into the model itself.

From a legal standpoint, this looks very much like a voluntary disclosure to a third party, and it makes a loss or waiver of privilege far more likely. Whilst unlikely that the information would be surfaced to other users, the mere fact of the disclosure may mean that information needs to be disclosed in court proceeding. Enterprise or paid versions of most tools offer stronger protections (for example, commitments not to train on your inputs), if selected, but the safe assumption for any free or consumer-facing tool is that your information is not private. When using enterprise or paid versions it is essential that you review the terms and ensure the appropriate settings are selected as they often default to permit sharing.

It is worth noting that the risk is not limited to privilege alone. Sharing confidential information with an AI provider may also breach your wider confidentiality obligations and your duties under data protection legislation such as the UK GDPR in relation to personal data.

Practical Guidance: Dos and Don’ts

Do:

• Assume that anything you put into a consumer AI tool could become public or be seen by others.
• Keep privileged or case-related material out of consumer AI tools.
• Remove or anonymise any personal data before you input into consumer AI tools.
• Use only tools or accounts that you or your organisation has approved for handling confidential and privileged information.
• Check the tool’s terms of service before use selecting the appropriate settings to restrict use of your data.

Don’t:

• Don’t paste your solicitor’s advice, correspondence about your matter, witness accounts or sensitive documents into a consumer AI tool.
• Don’t assume that pressing “delete” removes the data — it may already have been stored or used for training.

How to Check Whether a Tool Trains on Your Inputs

• Look in the provider’s Terms of Service, Privacy Policy and any “data controls” or “settings” pages for wording about using your data, inputs or conversations to “train”, “improve” or “develop” their models or services.
• Check whether there is an option to turn off training or chat history. Some tools allow you to opt out, and business or enterprise versions may commit not to train on your data.
• If the terms are unclear or you cannot find a clear commitment that reads “we do not train on your inputs”, treat the tool as unsafe for any confidential legal information.

When to Seek Guidance from Your Solicitor

• Before using any AI tool in connection with a live legal matter, dispute, investigation or regulatory issue.
• Before uploading, summarising or analysing documents connected to your case using an AI tool.
• If you are unsure whether information is privileged or confidential — ask first. It is much easier to protect privilege than to recover it once lost.

Conclusion: When in Doubt, Leave It Out

AI can be an incredibly useful tool in many areas of work and daily life. However, privileged and confidential legal information needs special care. The key message is straightforward: keep legal matters out of consumer AI tools — especially those that may train on your inputs — and always check the terms before you type.

When in doubt, leave it out — and ask your solicitor.

If you have any questions about legal privilege, the use of AI tools in your matter, please speak to your usual Ashtons contact who will be pleased to help.