The European Union cracks down on ChatGPT as the AI model fails to meet GDPR standards, with the European Data Protection Board’s report revealing insufficient compliance and ongoing data privacy issues with OpenAI’s attempts to align ChatGPT with EU regulations.
Even while the evaluation is still underway, OpenAI doesn’t seem to have advanced much since 2023.
The first progress report from the “GPT taskforce” was released by the European Data Protection Board (EDPB), and the news is not positive for ChatGPT.
The EDPB acknowledged but eventually ruled insufficient OpenAI’s attempts to align its core AI model, ChatGPT, with European Union regulations, particularly the extensive General Data Protection Regulation (GDPR).
The EDPB document states:
“Although the measures taken in order to comply with the transparency principle are beneficial to avoid misinterpretation of the output of ChatGPT, they are not sufficient to comply.”
The results coincide with OpenAI’s experience of handling temporary stop orders from many European member states over most of 2024.
Despite being warned and then prohibited in March 2023, ChatGPT and OpenAI were still in violation of Italian and EU data privacy rules, as Cointelegraph reported back in January. This information was provided by Italy’s data protection office.
The EDPB investigation claims that OpenAI hasn’t taken sufficient action in the interim to align ChatGPT with EU regulations.
The main grievance seems to be that ChatGPT frequently provides false information.
“As a matter of fact,” writes the EDPB, “due to the probabilistic nature of the system, the current training approach leads to a model which may also produce biased or made-up outputs.”
Concerns about “the outputs provided by ChatGPT are likely to be taken as factually accurate by end users,” whether or not they are, are also expressed in the report by the EDPB.
It’s unclear at this moment exactly how OpenAI could bring ChatGPT into compliance. For instance, the GPT-4 model has almost a trillion parameters and billions of data points.
By GDPR criteria, it would be impossible for people to go through the dataset and confirm its accuracy to the extent that it could be considered reasonably accurate.
Regretfully, the EDPB said clearly that “technical impossibility cannot be invoked to justify non-compliance with these requirements,” which is bad news for OpenAI.
Recently, GreatGameIndia reported that OpenAI ChatGPT decided to drop the AI voice mimicking Scarlett Johansson due to likeness issues and clarified that the voices in ChatGPT are carefully selected through an extensive process.
