As artificial intelligence becomes increasingly embedded in business operations, its legal footprint is growing just as fast. When it comes to intellectual property law, that footprint isn’t fully mapped out yet.
Although AI as a technology has been developing for years, local attorneys said many of the legal questions surrounding its use are still new.
“I don’t have 10 years of court decisions to look at to help me decide whether or not something is legal,” said Derek Stettner, IP attorney and partner at Milwaukee-based law firm Michael Best.
The courts have not yet ruled decisively on many AI-IP disputes, largely because litigation moves slowly by design. Local attorneys anticipate that some of the current high-profile lawsuits – such as the New York Times’ case against OpenAI – could go all the way to the U.S. Supreme Court. Until then, it’s not entirely clear how existing IP laws could be stretched and what the boundaries of AI are.
Still, attorneys agree that AI will not fundamentally rewrite IP law and that it will be largely worked into existing IP frameworks like previous technological advancements have been.
“AI is a transformative technology, and the legal landscape is adapting to its usage,” said Jon Stone, partner at Quarles & Brady and co-chair of the Milwaukee-based firm’s Artificial Intelligence Group. “In many ways, I view this sort of similar to the heyday of the Internet.”
[gallery columns="2" size="full" ids="615018,615017"]
A new frontier for copyright law
AI’s biggest legal collision point is with copyright law, attorneys agree.
Businesses looking to protect their work should be aware that their data could be used to train AI, some of which could be copyrighted material, though it’s unclear if that actually violates laws.
“When (AI) goes out and just grabs data, you scrape it off of someone’s website, you scrape it off of social media, what are you capturing?” Stettner said. “You’re potentially capturing copyrighted material.”
AI companies like OpenAI, Anthropic and Meta have trained their models on vast quantities of publicly available data – including articles, images, code, website content and other materials found online. Creators of some of this content argue that its usage amounts to unauthorized copying.
At the core of the debate is what it actually means for an AI model to “ingest” data. Creators argue that AI is copying their material to train itself and create new content, while AI companies counter that their use is more “transformative” than duplicative.
Businesses looking to incorporate AI into their workflows probably don’t need to worry about accidentally creating plagiarized or copywritten material, as long as they use AI responsibly.
“When you prompt AI to write something for you, it’s probably not going to deliver an exact copy of somebody else’s work unless you have a very detailed prompt,” Stettner said.
Still, attorneys urge companies to be cautious about making sure they don’t accidentally violate copyright laws by using AI-generated content.
“Basically, AI poses the same risks as a human,” said Dillon Durnford, an attorney and partner at Milwaukee-based Andrus Intellectual Property Law. “A human could plagiarize, so could AI. If you’re producing some sort of content for a website or a picture or something that includes someone else’s work, you might not know it.”
Less impact on patents
Compared to copyright, patent law will see fewer challenges in the face of AI, attorneys predict. That’s because patent infringement tends to involve the unauthorized making, using or selling of a protected invention, and existing legal tools already address those behaviors.
“If you’re in the industry of making and selling physical goods, that probably wouldn’t have as much of an impact,” Durnford said. “Even if somebody were to use AI to make a copy of your product and try to sell it as a knockoff or something like that, there’s traditional remedies for that.”
One way AI could change the patent landscape is by accelerating design-around strategies, in which companies look for ways to innovate around existing patents, according to Stone.
“Say I’m a pharmaceutical company and I have a medication on the market that took me years and years to find the compound that’s going to target this in this particular physiological pathway,” Stone said. “Somebody could now use an AI model to just churn through the drug discovery process and find compounds that are similar and be able to work around the patents I might have on that.”
While AI may provide a new tool for getting around existing patents, that practice is longstanding, Stone said, echoing Durnford’s point that AI largely presents the same IP risks that a human does.
“It’s an issue that we would have had to face regardless,” Stone said. “Competitors are always trying to work around other companies’ patents. It’s just that now AI might be a tool that can help them do that a little more efficiently.”
On the invention end, as U.S. law is currently written, AI cannot be an inventor of a patentable invention. Similarly, a federal circuit court recently ruled that simply applying an AI model to a new environment is not patentable.
“These large language models, most generative AI, they’re making predictive guesses of just what words should come next in the sequence of words within this context that it’s been given,” Stone said. “They don’t have invention capabilities yet, so I wouldn’t worry about an AI alone coming up with a patentable invention.”
Assuming AI continues to evolve and become more pervasive, as predicted, Stone said it is advisable to consider AI as an inventor – for better or for worse – in a company’s IP strategy.
“AI is not so disruptive that business leaders need to completely rethink their IP strategy, but it’s something to be mindful of and you want to be on your toes about it,” Stone said.
Practical takeaways for business leaders
Review and update your website’s terms and conditions
If your company publishes online content such as blog posts, product descriptions, photography or quarterly reports, AI models can easily access this data.
“If you’re worried about AI scraping everything on your website, there are some existing tactics you can use,” Stone said.
What to do:
- Have enforceable terms and conditions on your site that users need to accept or decline.
- Use technical controls like CAPTCHAs or robots.txt files to deter or slow down automated data scraping.
- Make sure proprietary data is not on your site, including old webpages that may not be linked to your homepage anymore.
Be cautious with what you tell AI
Attorneys stressed the risk of exposing company secrets inadvertently.
“If you’re using publicly available tools, any information that you put into them is not kept confidential,” Stettner said. “It’s possible to lose some of your trade secrets.”
What to do:
- Seek out private AI platforms over publicly available tools, like ChatGPT, for uses that may include proprietary data or content.
- Read the fine print of both public and private AI tools to understand its data retention clauses.
- If using a private AI tool, be thorough with questioning the AI vendor about its privacy, for example if its cloud is private or accessible to other AIs.
Avoid blind use of AI in content creation
If you’re using AI to create content for your business, like marketing copy or product descriptions, don’t assume the output is legally clean.
What to do:
- Maintain human oversight in any creative workflow that includes AI.
- Review imagery for unintended trademarks or brand elements that might infringe.
