AI and your IP
What You May Not Know About Intellectual Property And AI
Every time someone in your business types into an online AI tool, your information goes to another company. What happens to it next depends on that company’s terms, its settings, its staff, the courts, and whatever it decides to change later.
Where your data goes
What happens to what you type
Online AI services do not all treat your data the same way, and the differences matter.
- Free and personal plans can train on it. The consumer versions of the major assistants may use your conversations to improve their models, depending on a setting each user controls.[2][3][4] Business and API plans from the same companies generally do not train on customer data by default.[1]
- Your staff may not be on the business plan. In a 2025 survey of 1,000 US employees of large companies who use AI assistants at work, 57% admitted entering sensitive information into them.[5]
- Not training is not the same as not keeping. Even when data is not used for training, providers keep it for a period for abuse monitoring, commonly up to 30 days, and longer when it is flagged. One provider keeps flagged conversations for up to two years; another keeps chats that its human reviewers have read for up to three years.[1][6][2]
- Courts can order it kept. In May 2025 a US federal court ordered OpenAI to preserve ChatGPT output logs it would otherwise have deleted, as part of the New York Times copyright case. The order ended in September 2025, with some exceptions.[7]
- The terms can change. In 2025 and 2026 several well-known providers changed how consumer data may be used for training, and in 2023 Zoom reversed a terms update after customers objected to AI training on their content.[4][8][9]
None of this means a provider is acting in bad faith. It means that once your information leaves your network, its protection rests on another company’s promise, settings and circumstances rather than on your own controls.
Why it matters
What this means for your intellectual property
Trade secrets
Trade-secret law protects information only if you take reasonable measures to keep it secret. In a 2026 federal case, a court dismissed a trade-secret claim in part because using an online AI tool meant voluntarily sharing the information with the provider.[10]
Patents and inventions
The US Patent and Trademark Office has warned that information entered into AI tools may be retained by the vendor or used for training.[11] No court has yet decided that typing an invention into an AI tool is a public disclosure, but patent attorneys treat it as a real risk, especially in countries with no grace period for prior disclosure.
Confidential and privileged material
In 2026 a federal court in New York held that a defendant’s conversations with a consumer AI tool were not protected by attorney-client privilege. Anything typed into such a tool may have to be handed over in a legal case.[12]
Copyright and trademarks
Many AI models were built from material their makers did not own. One AI company agreed to a $1.5 billion settlement with authors over pirated books used in training, and the New York Times case against OpenAI and Microsoft is still in court.[13][14] The legal ground under these tools is still moving.
The alternative
AI that runs on your own hardware
Capable AI models can now run entirely on equipment you own. Several are published under the permissive Apache 2.0 license, including OpenAI’s gpt-oss, Mistral 3 and Google’s Gemma 4. The smaller gpt-oss model runs within 16 GB of memory; the larger one fits on a single 80 GB graphics card.[15][16][17]
When the model runs inside your building:
- Your documents, designs, code and customer records stay on your own network.
- No outside company keeps logs of what your staff ask, so its reviewers have nothing to read. Records on your own systems are still subject to your own legal obligations; the difference is that you hold them and you decide how long to keep them.
- A provider changing its terms does not change what happens to your data.
- It keeps working when the internet does not.
It is not magic. A local system still needs good security, the license terms of the model you choose still apply, and the very largest online models can be stronger on some tasks. For most business work with sensitive information, the trade is worth making. This is the work we do: private AI knowledge assistants and self-hosted automation that run on hardware you own.
Who We Are
White Rock Future Tech LLC provides AI-related services for small to large businesses. We also offer CIO/CTO services on a part-time and consulting basis. Our company uses AI in our daily work, which is kept in house on private networks when dealing with client materials and IP. We check each claim from our AI agents against their citations as well as checking for supporting information.
AI is a potentially game-changing tool for business today. We believe that the use of AI, in the correct hands and with appropriate guardrails, boundaries and rules will be as transformative as the introduction of the Personal Computer (PC) was back in the late ’70s and early ’80s. We can help you to put those guardrails and rules in place for your company.
Sources
Checked October 13, 2026. Provider terms change; follow the links for the current wording. This page is general information, not legal advice.
- OpenAI, “Data controls in the OpenAI platform”
- Google, “Gemini Apps Privacy Hub”, updated October 2026
- Microsoft, “Privacy FAQ for Microsoft Copilot”
- Anthropic, “Updates to Consumer Terms and Privacy Policy”, August 2025
- TELUS Digital survey of 1,000 US employees of large companies who use AI at work, February 2025
- Anthropic Privacy Center, “How long do you store my data?”
- Engadget, on the New York Times v. OpenAI preservation order, October 2025
- GitHub, “Updates to GitHub Copilot interaction data usage policy”, March 2026
- Variety, on Zoom’s AI training terms, August 2023
- Foley Hoag, “Litigating trade secret claims focused on generative AI”, May 2026
- USPTO, “Guidance on Use of Artificial Intelligence-Based Tools in Practice Before the USPTO”, 89 FR 25609, April 2024
- Proskauer, on United States v. Heppner (S.D.N.Y.), 2026
- JURIST, on the Bartz v. Anthropic settlement, July 2026
- Court docket, In re OpenAI, Inc. Copyright Infringement Litigation
- OpenAI gpt-oss model card (Apache 2.0)
- Mistral AI, “Mistral 3” (Apache 2.0)
- Google, “Gemma 4” (Apache 2.0)
Find out what your project really takes.
Book a free 30-minute consultation. You will leave with a realistic sense of size, cost and risk, whether or not we work together.