Most people recognize that generative AI, the use of artificial intelligence to generate content, represents an incredible tool. It allows companies and individuals the ability to automate certain processes, use a powerful computing tool to perform tasks, generate text, computer code, etc.

However, as recently discovered, using these tools incorrectly, or without proper review, can cause trouble. Recently in the news, lawyers representing the plaintiff in Mata v. Avianca, 1:22-cv-01461, SDNY, filed a brief with the court that included citations to non-existent opinions. The defendant’s attorneys filed a reply brief that informed the court and the attorneys that the opinions did not exist. Part of the attorneys’ troubles lies in their behavior after the opposing counsel highlighted the fake citations. Judge P. Kevin Castel called the attorneys to as show cause hearing as to why he should not issue a written decision and refer them to the New York bar for disciplinary action.

Other issues, not necessarily related to fictionalized content, may also arise from use of generative AI tools, and from other’s use of them. Use the tools, it may become imperative to use them to stay competitive against others that use them, but make sure you stay on top of it.

Make sure you review the results and ensure that the content does not violate others’ rights. Creation of content, text and otherwise, typically involves two parts – first, generate the content, then review the content before using it. Generative AI can help with the first part, but do not neglect the second. The below list is not exhaustive but provides examples of issues that may arise.

Check to make sure the content does not violate OTHER PEOPLE’S RIGHTS.

• This may include ensuring that the tool did not use text or images copyrighted by others. These include images or text owned by others in the output.
• Watch for open-source licensing violations. If using a tool to generate computer code, you can run a search on the code contents to ensure that it is not subject to an open-source license. Yes, open-source code is free, but may have attribution requirements that some of these tools do not meet. Some open-source code may also have use limitations, which these tools do not track. To gain further insight on pending litigation about these issues, click here.
• Privacy and data security violations. Google tried to launch Bard, their version of a chat bot, in the EU, but regulators there said, “not so fast.” The regulators have questions as to how they will protect consumers’ data and privacy.
• Violations of professional and fiduciary responsibilities. See above – enough said.

Check to make sure the content does not weaken YOUR RIGHTS.

• Sensitive information disclosure, including trade secrets, confidential information, client privileged information. Any prompts you give these chat bots become part of the chat bot and publicly available.
• You do not own the output. You cannot copyright something you did not create. You cannot therefore copyright or protect the output. It is not yours. For more information on copyright laws and AI, click here.
• Contracts generated by AI may have issues with over promising, establishing unfortunate precedents, and law violations.
• You may fall victim to the same violations for which you need to watch above.

Finally, the truth will out. Professors have tools available to them to detect plagiarism, such as Originality.AI. People who work in these fields may recognize chat box text. For example, a person who works in natural language processing received messages from a business contact trying to establish a relationship. The recipient recognized the messages as coming from the use of a large-language model (LL) chat bot, such as ChatGPT. The recipient declined to engage and the sender lost an opportunity and tarnished their reputation.

Even if you do not violate other’s rights, or diminish your own, without some care, you may damage your or your company’s reputation.

[View source.]