Everyone is still trying to figure out what artificial intelligence (AI) can and can’t do. Whether large language models (LLMs) like Gemini and ChatGPT will significantly change the world or remain useful AI tools is yet to be determined. A recent update to OpenAI’s usage policies suggests that its generative AI products may be able to offer a figurative nudge and a wink with their responses to user requests.
The policy update from October 29, 2025, appears to offer new guardrails for using AI-generated results. These include requirements for following usage policies when using OpenAI products. In addition to new safety and privacy provisions, OpenAI’s policy change has mandated that those who use ChatGPT or Sora for “high-stakes” decisions in certain topics must make the results subject to human review. Not surprisingly, legal concepts are listed among those that should be checked by a licensed attorney or another type of legal expert.
Does that mean OpenAI, which is partially owned by Microsoft, will stop providing legal advice and drafting templates of legal documents? Yes and no. The LLMs will continue to provide what users ask for, but results may come with a disclaimer stating that those who use AI should have their work checked by a licensed professional. The shared goal is to provide protection, with accurate results for the user and coverage for OpenAI, obtained through user awareness and consent.
Maybe Don’t Submit Your AI-Generated Court Filing Just Yet
AI wears many different hats, with usage including chatbots, music preference algorithms, and writing assistance programs. While new uses continue to be discovered, so too do potential limitations and implementations that raise moral, ethical, and legal red flags. Academic abuse of AI continues to be an issue, and hallucinations, courtroom missteps, and permissible usage remain ongoing concerns for legal professionals at all levels.
LLMs are designed to gather and learn from as much material as possible through a method called scraping, which is shrouded in controversy. Responding to prompts and requests from users, generative AI accesses its vast trove of collective data in an attempt to provide a proper response. Given the popularity and high levels of activity with their AI chatbots and generative AI programs, the retooling of OpenAI’s usage policies reflects the need to strengthen its own legal protection.
While the previous version stated that “we believe you should have the flexibility to use our services as you see fit,” OpenAI’s new policies “set a reasonable bar for acceptable use,” indicating a higher level of safety and control. This includes requiring professional advice to examine AI-generated results on topics such as medical advice, financial advice, health advice, and, of course, legal matters.
There’s a Reason for Law School
OpenAI’s new rules don’t mean that AI assistance won’t be available for medical information, healthcare questions, or any other topic a user wants to inquire about. They do make it clear that the results may be useful, but many subjects should be further checked by an expert in the field. This includes any and all legal document creation or advice-seeking.
To see if ChatGPT would still provide documents with a legal basis, we tasked OpenAI.com with creating a contract between a writer and an agent. It provided the request with the following caveat in the preface: “Please note that this is a general template, and it's recommended to consult a legal professional for jurisdiction-specific language and requirements.”
Not exactly a demand to “get thee to a solicitor,” but it does point the user in the direction specified by the updated version of the usage policies. It’s unlikely to eradicate questionable attempts to use AI-generated legal documents, but every little bit helps.
Related Resources
- District Judge’s Order Reeks of AI (FindLaw’s Practice of Law)
- Does the Slowdown of AI Adoption by Large Companies Indicate Changing Attitudes? (FindLaw’s Practice of Law)
- What Are the Elements of Negligence? (FindLaw’s Learn About the Law)
