In the three months since ChatGPT’s November 30, 2022 release, inaccuracies about the product have been swarming throughout the solo and small law firm community. Common misconceptions include the mistaken belief that ChatGPT scrapes web content, or that OpenAI, ChatGPT’s parent owns all copyright to user input and resulting output. Many of these myths arise from reliance on outdated material: the tech is changing so very quickly that a month-old article about ChatGPT may already be stale. And most lawyers are often too busy to poke around for the most up-to-date information.
In the hopes of setting the record straight, here’s a series of FAQs on ChatGPT for solo and small law firms that address some of solo and small firm lawyers’ most common questions about the technology. I can’t predict how long this post itself will remain current – but by including links to original source material, it should be easier for readers to check the accuracy.
FAQ: Will I be liable for plagiarism or copyright infringement by using ChatGPT since it scrapes content from other websites and public documents?
This question reflects a profound misunderstanding of how ChatGPT works. ChatGPT is known as a large language model tool — which means that it is trained and powered by mountains of data and computing techniques to make predictions to string words together in a meaningful way. Because ChatGPT does not scrape or copy content but instead relies on multiple sources, the resulting output is unlikely to carbon-copy any single existing work. As such, a copyright infringement on a prior work would not occur and if asserted, would be extremely difficult to prove.
For lawyers seeking more assurance against infringement, there are a several solutions. First, you could run your ChatGPT results through a web search engine or plagiarism detector to see whether it’s dangerously similar to prior works. Second, instead of relying on a “cut and paste” of ChatGPT output, put your own spin on it to make it even more original which would insulate you from copyright infringement claims.
FAQ: Can I prevent others from copying ChatGPT output that I use for my blog posts or marketing content?
That’s a topic that’s still in flux. In recent Guidance issued March 16, 2023, the Copyright Office reaffirmed longstanding policy that “to qualify as a work of ‘authorship’ a work must be created by a human being” and that it “will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.” That means that an article generated by a generalized ChatGPT prompt of “draft an article about five things not to do during a traffic stop” probably wouldn’t qualify for copyright protection.
But, the new Guidance acknowledges that “the Office will consider whether the AI contributions are the result of “mechanical reproduction” or an author’s “own original mental conception, to which [the author] gave visible form.” The answer will depend on the circumstances, particularly how the AI tool operates and how it was used to create the final work.” Under this emerging case-by-case inquiry, output produced by a cleverly worded prompt, or series of iterative prompts could theoretically qualify for coverage (though of course, you’d need to show proof of the creative process).
Still, the test for AI-enabled works is evolving and it’s not clear how far the Copyright Office will depart from original policy requiring human authorship. In the meantime, if you want to be assured of copyright protection, you’re best off avoiding a cut and paste and instead, adding your own personal spin to any ChatGPT output.
FAQ: Will Google penalize my site for using ChatGPT generated content?
In general, Google treats AI-generated content like ChatGPT no differently than any other user generated content — meaning, that Google elevates unique valuable content and demotes or de-lists garbage content and spam. Google confirmed this practice in Guidance dated February 8, 2023 where it emphasized that “Using automation—including AI—to generate content with the primary purpose of manipulating ranking in search results is a violation of our spam policies.” That said, Google acknowledged that many types of AI generated content can also be helpful to readers and would not be penalized in rank. Bottom line: if you unleash a torrent of a dozen keyword-laden posts every day to elevate SEO, Google’s search engine will respond negatively whether those posts were produced by ChatGPT or written by an actual person.
FAQ: Do ChatGPT’s terms of service assert copyright in all output and forbid users from referencing ChatGPT as a source?
ChatGPT use is governed by the terms of service for OpenAI, the company that developed Chat GPT. An older version of OpenAI’s terms of service included more restrictive provisions such as a claim of copyright entitlement to the output. The update terms assign any copyrights that OpenAI may have to the user, The new terms also warn users that non-API content provided can be used to train ChatGPT and allows for an opt out. The bottom line: terms of service for new products are always in flux or subject to change, so always check with the source (i.e., the actual terms) to make sure you’re dealing with the current version.
FAQ: Does using Chat GPT violate client confidentiality? As just noted, OpenAI does not use API content for training ChatGPT and allows an opt-out for non-API content. Which means that wide circulation of your input won’t be incorporated into future output. Even so, it’s prudent to treat ChatGPT as you would when seeking advice on a listserv or in a Facebook group. Generally speaking, an online inquiry about a case that doesn’t reveal any details about a client (such as asking about remedies for a parent who violates terms of a custody agreement by taking the child out of state) doesn’t breach attorney-client privilege. By contrast, posting your client’s social security number in an online forum and asking if anyone can run a credit check is a hard-stop no. In short, if there are details that you’re forbidden from posting online or sharing outside of the attorney-client relationship, don’t input them into ChatGPT. As with all new technology, common sense still applies.
FAQ: Can I use the results of ChatGPT and court filings?
You can – but if you don’t exercise some due diligence, you put yourself at risk. While a strong starting point for research and issue-spotting, ChatGPT results may not always be accurate enough to plop right into a brief without doing some follow up research. But realize, this isn’t a flaw that’s unique to ChatGPT. After all, you wouldn’t string cite case annotations from the most reputable legal treatise without first reading the cases, so why would you simply cut and paste ChatGPT output into a legal brief of memo. If there’s one thing that law school actually did teach us, it’s that it’s never a sound practice to use any secondary legal authority, no matter the source, without additional research or review.
FAQ: Do ethics rules foreclose use of Chat GPT?
No. In fact, ethics provide some rough guidance on responsible use of new technologies like ChatGPT. Under ABA Model Rule 1.1 and states that have adopted it, lawyers have an ethical obligation to stay abreast of the risks and benefits of emerging technology. That means that lawyers are obligated to take proactive steps to learn about new tools like ChatGPT that can help clients. ABA Ethics Opinion 08-451 also allows lawyers to outsource work to nonlawyers (including automated or tech solutions) provided that they adequately supervise the work and protect confidential information. This reinforces the importance of checking ChatGPT’s work before using it.
FAQ: How can I keep up with all these changes in ChatGPT?
For starters, follow MyShingle.com and other blogs that track copyright law or legal tech. And if you’re looking for regular tips and tricks on practical ways to incorporate ChatGPT into your practice, check out law professor Josh Kubicki’s Brainyacts newsletter.