Can AI be sued? Here are 5 potential legal issues

Mar 24, 2023

Simone Del Rosario Business Correspondent

By Simone Del Rosario (Business Correspondent), Emma Stoltzfus (Editor), Brent Jabbour (Producer)

Artificial intelligence is already the next big thing, from OpenAI’s ChatGPT to Google’s Bard AI. But as generative AI platforms increasingly get in the hands of everyday people, could the technology end up on the wrong side of a lawsuit? Here is the legal action threatening AI in this week’s Five For Friday.

#5: Defamation

Chatbots have a tendency to make up facts. They are trained with a ton of data, but when given prompts, bots like ChatGPT don’t scrub the internet for new information. As a result, the AI sometimes convincingly spews sentences out of thin air, including quotes. This has journalists, lawyers and academics chatting about whether AI chatbots can be guilty of defamation. It would be pretty tough to prove a computer acted with malicious intent in spreading lies, but will the developers behind the AI platforms be held accountable?

#4: Intellectual Property Infringement

Character.AI lets users chat with their favorite fictional characters and real-life celebrities. Generative AI chatbots like this use available data in the ether to figure out how someone might respond. It can emulate text-based conversations with the likes of Kanye West or Keanu Reeves, among others. But it gets a little more legally murky when it offers the chance to converse with fictional characters like Luke Skywalker or Spider-Man. Is Disney going to continue to allow Character.AI, which is valued around $1 billion, to use its intellectual property, especially when it has no control over what is said?

#3: Deepfake

Machine learning has made way for deepfake technology, which allows users to take an existing video, replace the face and even replicate someone’s voice. These have been making waves on social media with videos featuring Tom Cruise and a number of politicians. But there are even more nefarious actors out there who use the AI to make pornographic material. As of right now, few states have laws on the books that specifically address the technology. In the week of March 19, 2023, TikTok said it won’t allow deepfakes of any “real private figure,” but it appears public personalities are still fair game, as long as it’s disclosed and they aren’t making any endorsements.

#2: Malpractice

Is it possible for AI to be guilty of malpractice? That’s a question being asked as part of a proposed class action lawsuit against DoNotPay. The company made headlines when it offered $1 million to anyone willing to use its “robot lawyer” to argue a case in front of the Supreme Court. Straight Arrow News even talked with DoNotPay CEO Josh Browder after he made the offer. But then he said he received “threats from state bar prosecutors” when he tried to use the AI lawyer in traffic court. The class action suit stems from a man that used the online service for a number of legal documents, claiming he received “substandard and poorly done results.” Concerns over AI malpractice can only get worse as people use the tech for medical advice, and who knows where it goes if AI therapy takes off.

#1: Copyright

Generative AI has to be fed a ton of information to be able to respond to user input and this can be a copyright law minefield. Getty Images is already suing Stability over its art crafted by AI Stable Diffusion. Getty’s suit claims the company scraped 12 million of its high-quality imagery to train the AI. Getty sells those images at a premium and has already made licensing deals to a number of AI companies. Artists also sued Midjourney, a similar AI platform that was used to generate images of every American president, but they look “cool” and have a mullet. Take a look at all of that glory instead of focusing on the people out there making fake AI images of former President Donald Trump being arrested.

Tags: Artificial Intelligence, ChatGPT, Deepfakes, Defamation, Midjourney

Simone Del Rosario:

FROM CHATGPT TO BARD, ARTIFICIAL INTELLIGENCE IS ALREADY THE NEXT BIG THING. BUT COULD IT BE ON THE RECEIVING END OF THE NEXT BIG LAWSUIT? HERE’S THE LEGAL ACTION THREATENING A-I IN THIS WEEK’S FIVE FOR FRIDAY.

WE KNOW FROM PLAYING AROUND WITH THESE CHATBOTS THAT OFTEN, THEY MAKE UP STUFF, OUT OF THIN AIR IT SEEMS, INCLUDING QUOTES. THIS HAS JOURNALISTS, LAWYERS AND ACADEMICS ALL CHATTING ABOUT WHETHER CHATBOTS CAN BE GUILTY OF DEFAMATION. IT’D BE PRETTY HARD TO PROVE A COMPUTER ACTED WITH MALICIOUS INTENT IN SPREADING LIES, BUT WILL A-I MAKERS BE HELD TO ACCOUNT? TBD.

CHARACTER A-I LETS YOU CHAT WITH YOUR FAVORITE FICTIONAL CHARACTERS AND REAL LIFE CELEBRITIES. USING GENERATIVE A-I, CHATBOTS LIKE THIS USE AVAILABLE DATA IN THE ETHER TO FIGURE OUT HOW SOMEONE *MIGHT RESPOND, THEN PUTS THOSE WORDS RIGHT IN THEIR MOUTH. YOU GOTTA WONDER IF DISNEY IS COOL WITH THEM USING THEIR IP ESPECIALLY SINCE CHARACTER AI IS VALUED AROUND A BILLION DOLLARS.

YOU CAN’T EVEN TELL THAT’S NOT THE REAL TOM CRUISE! MACHINE LEARNING HAS MADE WAY FOR DEEP FAKES, WHERE YOU TAKE AN EXISTING VIDEO, REPLACE THE FACE AND COPY A CELEBRITY’S VOICE. YOU’VE EVEN GOT POLITICIANS OUT THERE SPOUTING NONSENSE. BUT THERE ARE EVEN MORE NEFARIOUS ACTORS MAKING, UH, EXPLICIT MATERIAL WITH DEEP FAKES… AND RIGHT NOW, ONLY A FEW STATES HAVE LAWS ON THE BOOKS TO ADDRESS IT. THIS WEEK TIKTOK SAID IT WON’T ALLOW DEEP FAKES OF ANY “REAL PRIVATE FIGURE,” BUT IT’S FAIR GAME ON PUBLIC PERSONALITIES, AS LONG AS IT’S DISCLOSED AND THEY’RE NOT MAKING ENDORSEMENTS.

CAN AI BE GUILTY OF MALPRACTICE? DONOTPAY MADE HEADLINES AFTER OFFERING A MILLION BUCKS TO ANYONE WILLING TO USE ITS ROBOT LAWYER IN FRONT OF THE SUPREME COURT. WE EVEN TALKED TO THE CEO AFTER HIS OFFER, BUT THEN HE GOT “THREATS FROM STATE BAR PROSECUTORS” WHEN HE TRIED TO USE THE ROBOT LAWYER IN TRAFFIC COURT. NOW IT’S FACING A PROPOSED CLASS ACTION SUIT FROM A GUY WHO USED THE SERVICE FOR A NUMBER OF LEGAL DOCUMENTS, CLAIMING HE GOT “SUBSTANDARD AND POORLY DONE RESULTS.” THIS CAN ONLY GET WORSE AS PEOPLE TRY TO RELY ON AI FOR MEDICAL ADVICE. HAVE WE TRIED IT YET FOR THERAPY?

FINALLY, THE COPYRIGHT CLAIMS. GENERATIVE A-I HAS TO BE FED A TON OF INFORMATION TO BE ABLE TO RESPOND TO USER INPUT. AND GETTY IMAGES IS SUING A COMPANY CALLED STABILITY OVER ITS ART A-I STABLE DIFFUSION. GETTY CLAIMS THE COMPANY SCRAPED 12 MILLION OF ITS HIGH QUALITY IMAGES TO TRAIN THE TECH. AND GETTY SELLS THOSE IMAGES AT A PREMIUM AND ALREADY LICENSES IT TO SOME A-I, SO YOU SEE THE CONFLICT HERE. OTHER ARTISTS ARE SUING MIDJOURNEY, A SIMILAR SERVICE THIS GUY USED TO CREATE EVERY AMERICAN PRESIDENT, BUT LOOKING COOL WITH A MULLET.

YOU KNOW WHAT, LET’S KEEP THOSE PRESIDENTS ROLLING. HAVE A GREAT WEEKEND, THAT’S FIVE FOR FRIDAY, I’M SIMONE DEL ROSARIO, IT’S JUST BUSINESS.