Legal Considerations and Best Practices for AI Assisted Software Development

To mitigate risks, companies should ensure meaningful human contributions to development, maintain detailed documentation, use enterprise AI platforms with proper contractual protections, and implement robust confidentiality measures.

Companies large and small are increasingly using generative AI platforms to develop software, often making the process quicker, easier and more cost-effective. But companies drawn to the promise of greater efficiency may overlook the substantial legal and commercial risks associated with AI-assisted software development. These risks, if not properly addressed, can jeopardize intellectual property rights and expose companies to liability.

This article examines the key intellectual property risks arising from the use of AI in software development—including limitations on copyright and patent protection, the potential loss of trade secret protection and exposure to intellectual property infringement claims—and offers best practices to help mitigate those risks.

Copyright and Patent Protection Limitations

U.S. copyright law protects only those elements of a work, such as a software program, that are authored by a human. The U.S. Copyright Office and federal courts have accordingly concluded that works created solely by generative AI are neither protectable nor registrable (Thaler v. Perlmutter, 687 F. Supp. 3d 140, 147 (D.D.C. 2023)). With respect to AI-assisted works, the Copyright Office has concluded that “given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output [and] essentially function as instructions that convey unprotectible ideas.” See U.S. Copyright Office, Copyright and Artificial Intelligence, Part 2: Copyrightability 18 (2025).

U.S. patent law similarly requires human inventorship. The USPTO and federal courts have affirmed that an inventor of a patent must be a natural person, so an AI system cannot be an inventor (Thaler v. Vidal, 43 F.4th 1207 (Fed. Cir. 2022)). However, following Thaler, the USPTO issued guidance explaining that, although inventions created solely by AI are not patentable, AI-assisted inventions may be patentable if a human significantly contributed to the claimed invention (89 Fed. Reg. 10,043 (Feb. 13, 2024)). The USPTO views AI as a tool “used by the human inventor who conceived the claimed invention” and states that the relevant inquiry is whether that person conceived the invention by having “a specific, settled idea, a particular solution to the problem at hand, [and] not just a general goal or research plan.” See "Revised Inventorship Guidance for AI-Assisted Inventions," 90 Fed. Reg. 54,636, 54,637 (Nov. 28, 2025).

Accordingly, although AI-assisted software development may be commercially valuable, the use of AI can create gaps in copyright and patent protection. Those gaps may expose companies to competitive risk by limiting their ability to prevent others from using or copying valuable software developed with AI assistance.

Best Practices to Maximize Copyright and Patent Protection

Given the current human-centric approach to copyright and patent protection, preserving meaningful human authorship and inventorship in software development is essential to maximizing such protection.

The following best practices can help enhance copyright and patent protectability of software developed with the assistance of AI:

1. Ensure that humans contribute substantively to software code, architecture, design and other protectable elements.
   
   Copyright and patent protection depend on human authorship and inventorship, respectively, so meaningful human contributions are key to maximizing protectability. Human involvement is essential in the ideation, review, editing and modification of AI-assisted software.
   
   Overreliance on AI-generated code with respect to core or strategically valuable functionality should be avoided because it will likely jeopardize the copyright and patent protectability of critical code, opening the door to competitors’ unfettered use of such code for their own benefit.
2. Maintain detailed documentation evidencing human contributions to the software development process.
   
   Track and document the specific ways in which humans and AI participated in the software development process. Evidence of human authorship or inventorship may be required in connection with copyright or patent registration and enforcement efforts.
3. Adopt a layered copyright registration strategy.
   
   In addition to registering the software as a whole with the Copyright Office (identifying only human contributors as “authors” and disclaiming AI-generated material where appropriate), consider separately registering the software’s human-authored components (e.g., code modules, audiovisual assets, documentation, interfaces, music, storyboards, designs or other elements that were authored by humans). This layered registration strategy creates a robust basis for enforcing your rights in the more clearly protectable (i.e., human-authored) components of your software.

Potential for Loss of Trade Secret Protection

The use of generative AI in software development also poses significant risks to trade secret protection. Prompting a publicly available generative AI platform with sensitive information may render that information generally known or readily ascertainable, thereby extinguishing its trade secret status.
For example, employees may disclose proprietary source code, algorithms or confidential business information when prompting AI systems. If the AI platform trains on that information and incorporates it into outputs provided to third parties, the information may no longer be a secret, and its trade secret status may therefore be compromised.

Trade Secret Protection and Confidentiality Measures

The following best practices can help preserve and utilize trade secret protection for software (or elements thereof) developed with the assistance of AI:

• Exercise caution with your inputs and use enterprise versions of AI platforms where appropriate.
  
  Do not input sensitive source code, proprietary algorithms, customer information or strategic business data into an AI platform absent adequate contractual protections. Many public or non-enterprise platform licenses allow AI providers to train their platforms on the data you input, and the platforms may disclose such data in their outputs to other users.
  
  Consequently, enterprise or other appropriately secured versions of AI platforms should be used when dealing with sensitive development work. Carefully evaluate the applicable license agreement to ensure that the AI provider is not permitted to use, retain or incorporate materials that you input into the platform except as necessary to provide services to your company.
• Adopt reasonable confidentiality and protection measures within your company.
  
  Within your company, maintain audit trails and restrict access to proprietary software and development environments to those with a specific need for access to minimize the risk of inadvertent trade secret disclosure through the use of AI platforms.
  
  Clearly mark sensitive materials with confidentiality legends and proprietary notices so that their confidential nature is apparent to those with access to them. Additionally, where practical, require employees, contractors, licensees and partners with access to sensitive materials to sign nondisclosure agreements that prohibit inputting such materials into unauthorized AI platforms.
• Utilize technological measures to prevent unauthorized use of AI-assisted software.
  
  When feasible, distribute AI-assisted software only in object code form (i.e., nonhuman-readable form) and consider using digital rights management technology that precludes or restricts unauthorized use.
• Carefully draft your software license agreement.
  
  The license grant in your software license agreement should clearly define the scope of permitted uses and any related restrictions. It should prohibit access or use by competitors, reverse engineering and attempts to access source code, as well as unauthorized copying, modification or redistribution.
  
  The license agreement should also contain strong confidentiality provisions that, among other things, designate the software (including its source code) as confidential and subject to trade secret protection, and expressly prohibit its use outside the scope of the license.

Intellectual Property Infringement Risks

Using AI platforms to assist in software development exposes companies to potential intellectual property infringement risk. A particularly challenging risk to manage arises from AI platform training. Many generative AI platforms are trained and fine-tuned on copyrighted works of authorship without the authorization of their owners. Numerous lawsuits allege that such unauthorized use constitutes copyright infringement. Consequently, companies that use these AI platforms to develop software may also face infringement claims if AI-generated outputs incorporated into their software reproduce, or are substantially similar to, protectable elements of third-party copyrighted works.

Mitigation of Intellectual Property Infringement Risk

Although the risk of infringement claims cannot be eliminated entirely, adopting the following best practices can help mitigate such risk:

• Conduct diligence and use caution when making certain representations.
  
  Conduct thorough diligence on AI-generated components of your software to identify potentially infringing elements. Additionally, when creating AI-assisted software for (or licensing it to) others, carefully consider the scope of any representation or warranty that the software does not infringe third-party intellectual property rights. Any such representation or warranty should include appropriate qualifications, disclaimers and limitations with respect to components developed using an AI platform.
• Address risk allocation in agreements related to AI-assisted software.
  
  Software-related agreements should address the allocation of risk associated with any AI-generated components of the software. In particular, consider whether any available protections or indemnities from the AI provider (particularly any contained in enterprise-level licenses) can be passed through to your counterparty. If not, consider whether and to what extent you should indemnify the counterparty in light of the pricing and other provisions of the agreement.

Conclusion

If not properly addressed, the risks associated with AI-assisted software development can have significant consequences, including forfeiture of intellectual property rights and exposure to intellectual property infringement claims. Other issues (e.g., cybersecurity, software integrity and bias) must also be considered. To minimize these risks and protect your company’s legal and commercial interests, it is critical that you consult with legal counsel experienced in addressing these issues throughout the AI-assisted software development process.

Reprinted with permission from the September 2 edition of the LegalTech News © 2026 ALM Global Properties, LLC, trading as Centellic. All rights reserved. Further duplication without permission is prohibited, contact 877-256-2472 or asset-and-logo-licensing@alm.com.