Organizations that don't adopt AI internally will put themselves on a path for extinction. The competitive pressures will simply become too large. I've seen companies go from shipping big projects in a quarter to shipping big products in a matter of weeks. The speed of creation and research enabled by AI can't be matched by humans alone.
However, AI presents many challenges when it comes to protecting your intellectual property. Material solely created by AI isn't eligible for copyright or patent protection in the US. Simply providing prompts to an AI isn't enough — the US Copyright Office requires that a human exercise creative control over the expressive elements of the work.
So how do you protect your product and other valuable creations? There are two strategies — and some practical advice on adoption.
Disclaimer: I am not a lawyer. You should consult with a qualified expert before making any decision regarding your IP. I make no warranties and disclaim all liability. Caveat Emptor.
Trade Secrets
Think of Coca-Cola's secret recipe. It's locked in a vault somewhere, and only a handful of employees have access to it. That's a trade secret. Although you can't keep IP protection and reveal a trade secret to the public, you can maintain the protection of that IP indefinitely. Copyrights expire (currently life of author plus 70 years, or for work made for hire, 95 years from publication or 120 years from creation, whichever is shorter).
So, even if you can't protect AI-generated content with copyright or patents, you can protect them as trade secrets. This means that you must take strict measures to prevent public disclosure of the secrets. You should have appropriate confidentiality agreements with any vendors who might have access to your content, including AI vendors.
Practically speaking, this means NDAs and IP assignment agreements with all employees and contractors. When it comes to AI tools, you will need to pay for any AI service you use, and also ensure that you do not allow the vendor to train off of your prompts or data. Don't let your employees put sensitive data into unapproved AI tools -- including Google search.
Disclosing your trade secrets to a free AI tool is a terrible idea. These tools are free to use your content to train their AI models: essentially you are publishing your IP in a textbook for training AI.
Track Humanity
You can copyright the human-contributed parts of a work. In a codebase consisting of AI-generated code and human-generated code, you can copyright the human-generated parts.
Ultimately, this means you need to keep good records of what the AI and what a human contributes to a codebase. If you don't, the US Copyright Office will reject your attempt at registration.
Both AI and version-control systems can help with this. For instance, Claude Code will automatically add the annotation "Co-Authored-By: Claude Opus 4.6 noreply@anthropic.com" to git commits it makes. If your employees add their human-generated code in separate commits, you will establish a record of authorship.
However, be careful not to squash AI and human commits together if you use this approach. Talk to your software engineers if that doesn't make sense to you.
Practical Considerations
In reality, I've observed very few technology businesses register software or hardware designs with the US Copyright Office. Marking copyright without registering it before infringement offers only limited protection. You can sue for damages and profit resulting from the violation, but not the statutory damages available to a work registered with the optimal timing.
This practice makes sense. In technology businesses, the larger concern is not stolen content. The big concern is surviving due diligence. Regardless of copyright, an investor wants to see that you have the right to use the IP you claim.
In reality, most technology businesses rely on secrecy to protect their internal IP. That's great news, because you can protect AI-generated work as a trade secret.
Publicly published content is a different story, of course. What you publish in a book or on your web page isn't a trade secret. It can't be. Without meaningful human authorship, published AI-generated content enters the public domain with no copyright fallback. If you want to prevent others from taking IP you plan to develop, you will need to establish copyright -- that is, you need to have a human write it, document the fact, and assign the rights to your business.
Patents are the one sticky area. If you have invented something novel and of value, you may wish to file a patent both as a moat around your business, and as an anchor securing your right to use the invention. A trade secret kept purely internal generally won't qualify as prior art to invalidate a competitor's later patent. There's one exception: if you commercially sell a product embodying the secret, that sale can trigger the on-sale bar under Helsinn Healthcare v. Teva (2019), even if the underlying details remain confidential.
There's also an important safety net: the AIA's prior use defense (35 U.S.C. § 273) allows a trade secret holder who commercially used an invention at least one year before a competitor's patent filing to continue that use — even if the competitor's patent is valid. This doesn't invalidate the patent, but it protects your right to keep doing what you were already doing.
For this reason, you should use caution when creating new inventions; don't introduce any doubt that it lacks a human inventor. And consider whether your invention is independently discoverable — if a competitor could reach it through their own R&D, patenting may be the stronger play.
Conclusions
Your best bet for protecting IP is to document what humans create and use trade secrets to protect everything else. Here are some strategic actions to take today:
- Have your general counsel or outside IP attorney conduct an IP audit; identify what content, code, and inventions are most valuable to your business
- Have your CTO or VP of Engineering audit AI usage across the organization, specifically where it intersects with your most valuable IP
- Have your CEO or COO work with legal and engineering to determine what policy and training you need to protect that IP as it interacts with AI tools
Tactically, have your engineering and legal teams ask of each tool:
- Do we have an agreement with this AI vendor that preserves the secrecy and ownership of our IP?
- Have we configured this tool to prevent it from using our data for training new models?
- Do we have a path to audit what AI versus a human is creating?
Finally, have your CTO periodically take stock of all software tools used in the organization. Which tools have added AI? Are those tools protecting your IP? What free tools might tempt staff to leak IP to an AI vendor?
The law will evolve — some AI companies will lobby Congress for stronger copyright protections, while others benefit from the status quo when training their models off of scraped data. Plenty of businesses will claim copyright over AI-generated content without doing the work to establish it. Whether competitors call that bluff remains to be seen. Regardless of which direction the law goes, the fundamentals above hold.
For a deeper dive into the case law and specific recommendations, see my full research report on AI-generated content and IP ownership in US law.