New troubles are coming for OpenAI, the company behind ChatGPT. Austrian activist Max Schrems, very active against the internet giants when it comes to personal data, has launched proceedings against the creator of the famous chatbot.
Have you been curious enough to ask ChatGPT what it knows about you? The results probably did not live up to your expectations. It's the same for Max Schrems. The Austrian also did the experiment, and the result obtained turned out to be off the mark. But, rather than moving on, he wanted to correct the inaccurate information.
Advertisement
This is where things got complicated. According to the person concerned, his requests for correction went unheeded. “ Although the complainant's date of birth provided by ChatGPT was incorrect, OpenAI refused his request to rectify or erase the data, arguing that it was impossible to correct the data ”, it is said.
But Max Schrems is not an Internet user like the others. He is also an activist who is at the head of an association very involved in privacy issues. Silicon Valley has learned over time to be wary of NOYB (None of Your Business), because the NGO's legal actions have shaken up the transatlantic framework on personal data several times.
ChatGPT says anything and OpenAI does not correct
And it is precisely before the Austrian body responsible for the protection of private life that Max Schrems intends to drag OpenAI. The American company, designer of the chatbot, is accused of not respecting certain provisions of the General Data Protection Regulation (GDPR), even though it has fully applied since 2018.
According to NOYB, companies like OpenAI “ are unable to ensure that chatbots (…) comply with EU law when processing data about individuals. » However, the GDPR requires that personal data be accurate. It's even an older rule, dating from 1995.
Advertisement
Gold, ” OpenAI openly admits that it is unable to correct incorrect information on ChatGPT », Points out the association. She also accuses him of not being able or knowing how to say “ where the data comes from or what data ChatGPT stores about individual people. » A situation deemed unacceptable, given the weight of ChatGPT.
A complaint was therefore filed, it was announced this Monday, April 29, to force OpenAI to give more satisfactory answers to Max Schrems, rather than refusing him. The activist also wishes to obtain the data concerning him, and the sources used by ChatGPT to construct his response. Here too, OpenAI was not cooperative.
When a chatbot starts talking nonsense, it is considered to be “hallucinating”. This is one of the great challenges of generative artificial intelligence, with respect for copyright (not all online texts are royalty-free), biases when training the chatbot and computer vulnerabilities.
“ Inventing false information is already problematic in itself. But, when it comes to false information about people, the consequences can be serious », Commented NOYB’s lawyer. For the association, it is time to force OpenAI to comply with the GDPR, and to consider paying a fine.
The problem raised by NOYB is not new. In fact, this defect in the processing of personal data had begun to arise in the months following the public availability of ChatGPT, in November 2022. From the spring of 2023, reproaches of this type began to arise, both in Italy as in France, as well as in the United States.