OpenAI, the company behind wildly popular generative AI chatbot ChatGPT, is being sued for $3 billion by a group of people who allege it stole “vast amounts” of personal data to help train its artificial intelligence models.
In a 157-page lawsuit that is seeking class action status in San Francisco District Court, the anonymous individuals claim that OpenAI violated privacy laws by using “personal information obtained without content” as part of a trawl of 300 billion words of internet content that has informed ChatGPT's knowledge base and responses.
The group adds that it is filing suit for potential damages on behalf of millions of individuals who have had their data used by the company without permission, including children. All in all, the lawsuit is a dramatic one that accuses ChatGPT owner OpenAI of nothing than less than risking “civilizational collapse” in its pursuit of profit.
OpenAI Accused of “Secret Scraping”
At the heart of the lawsuit is the accusation that OpenAI has been running a far-reaching web scraping program in secret as it looks to turn ChatGPT into not only the most advanced AI chatbot around, but the future of technology as a whole.
As first reported by Bloomberg, the plaintiffs claim that the company has violated numerous terms of service agreements, as well as as state and federal privacy and property laws, in running the operation to train ChatGPT. Two of the laws specifically mentioned as being breached are the Computer Fraud and Abuse Act, as well as the Electronic Communications Privacy Act. The suit is unflinching in the language it uses to describe OpenAI's practices, saying they amount to nothing less than “theft.”
Related: Does ChatGPT Save Data?
Even more specifically, it says that OpenAI has illegally accessed and misused personal data via its third-party integrations. This is said to include things like image and location data from Snapchat; Spotify music preferences; financial details from Stripe; and even private conversations taking place on Slack and Microsoft Teams. It doesn't stop there, though, adding that the personal hobbies, religious beliefs, political views, gender identity, and sexual preferences of millions have been integrated into ChatGPT without them knowing it.
The “AI Arms Race” Turns Nasty
As well as its fierce allegations of privacy-related crimes, the document levels what can only be described as personal accusations against OpenAI and its founders. It claims that the organization has torn up its original principles of developing AI in a way that will “likely benefit humanity as a whole” in favor of “winning the AI arms race” and the brazen pursuit of profit, contending that the firm is expected to make around $200 million this year.
The no-holds-barred lawsuit even goes as far as to name new OpenAI investor Microsoft – reportedly ploughing $10 billion in the AI company – as a co-defendant in the case. It's the latest piece of drama to engulf the ChatGPT maker, which finds itself the subject of intense regulatory debates stretching from Capitol Hill all the way to the European Union.
While generally light on specific instances of harm caused to individuals, one interesting thing that the lawsuit does make plain is that the extent of OpenAI's data usage meant it should have been formally registered as a data broker, as required by law. This is just one example of the company ignoring the legal obligations surrounding the acquisition and use of personal data, the suit adds.
Read More: Best ChatGPT Alternatives
What Could Happen to OpenAI in the ChatGPT Lawsuit?
Let's be clear: a lawsuit being filed against OpenAI is a pretty big deal, but there's no saying it will actually gain any traction in the courts. To our eye, it seems like the plaintiffs in this case have a whole lot to prove if they want to take down the ChatGPT creator, or force it to change how it operates.
Firstly, getting classified as a class action lawsuit is no given. The court in San Francisco will need to agree that the (anonymized) named plaintiffs share a common cause with the unnamed plaintiffs and have presented a case that's genuinely representative of this potentially huge group of people. Beyond this, it seems likely at some stage that more substantial proof of wrongdoing will need to be presented, specifically examples of how ChatGPT has actually produced content based on content or data acquired illegally or unfairly from the plaintiff group. This is no small task, either.
Finally, it's worth noting that the $3 billion figure attached to the lawsuit is almost certainly just a placeholder. Assuming all of the above were proven to the satisfaction of the courts, it would then be down to the lawyers to argue how much damage it equates to and why. It can be an uphill struggle proving such things, especially given that the presumption of innocence extends to lawsuit defendants, so right now it's little more than a headache for OpenAI's legal team and some more unwanted attention.