Using generative AI platforms: 10 best practices
AI-generated materials have the potential to offer enormous benefits to businesses, but also may pose significant potential legal, regulatory and reputational risks. Here we offer ten high-level guidelines to help you understand the principles underlying the use of generative AI platforms in your business.
1. Know your rights. Material created purely by AI is not protected by copyright. Creative additions to AI-created materials may be protected, but only the human additions. The human selection, coordination or arrangement of certain AI materials may also be protected. It is important to know and keep track of what was created by the AI and what was created or changed by humans.
2. Check their terms. Always check terms and conditions of the AI platform. Each platform has different terms which allow or prohibit its use for certain purposes. Always check to ensure that the platform permits your use (eg, commercial use, reproductions). These terms may require that you purchase and maintain a license to use the material.
3. Check your terms. When using generative AI for clients or to create materials for third parties, know the terms of your agreement with the third party. The agreement may restrict AI use. The agreement may require you to warrant your ability to assign copyright in your creation to the third party. You may not be able to make an unqualified warranty if generative AI is used to create work product.
4. AI Is the beginning, not the end. Generative AI is a great way to develop ideas at the beginning of the creative process. But it is best to exercise caution before using mainstream generative AI platforms to create final content for clients or for publication. More bespoke applications of generative AI may be utilized for specific company outputs following proper legal vetting.
5. Be original. Use original prompts and input data when using generative AI, rather than copying or adapting content from other sources.
6. Be safe. Refrain from using prompts that contain confidential or legally privileged information. If your AI source is secure (ie, internal to the company), feeding confidential data may be safer, but doing so may still violate contractual terms if the data belongs to a third party.
7. Be your own brand. Avoid using brand names, logos, or other trademarks in your prompts or input data. Doing so helps reduce the likelihood that the output will infringe on third-party trademarks.
8. Be specifically general. Specific prompts often generate useful end products. But using specifying material, including celebrity names, styles of art or design, specific images, and other material that identifies third-party material increases risk that your material will come into conflict with another’s rights.
9. Keep your records. Depending on your use case, it may make sense to keep detailed records of your input data and the output generated by your AI, as well as the changes you make to the material, in case you need to demonstrate the source of your work.
10. Call your lawyer. Should you have any doubts about the AI-generated material you are considering, ask your legal counsel to review them in final form.
To find out more about these best practices and our Artificial Intelligence and Data Analytics work, please contact Bennet Borden, Danny Tobey or Gina Durham, and visit our resource center, Focus on Artificial Intelligence.
Five key takeaways on recent US copyright policy developments for artificial intelligence
21 March 2023 .6 minute read
Text to image AI generators: familiar legal questions about this novel, fast-growing...
24 January 2023 .5 minute read