Hallucinating fake birthday
Data protection advocates are taking action against ChatGPT
April 29, 2024, 10:44 a.m
Listen to article
This audio version was artificially generated. More info | Send feedback
ChatGPT also spits out incorrect information, for example the date of birth of a well-known person. European data protection advocates want to stop this and accuse OpenAI of inaction. A fine should bring improvement.
The European data protection organization NOYB, together with an affected European citizen, has filed a complaint against the ChatGPT developer Open AI because the AI program invents “false information about people”. The Vienna-based organization None of Your Business (NOYB), co-founded by data protection activist Max Schrems, criticized: “ChatGPT is hallucinating and not even OpenAI can stop it.”
According to NOYB, the US company is aware of the problem. “The company openly admits that it is unable to correct incorrect information on ChatGPT,” NOYB said. “The company doesn’t even know where the data comes from or what data ChatGPT stores about individuals.”
This was unacceptable as it violated the European General Data Protection Regulation, it said. If a system cannot provide accurate and transparent results, it should not be used to generate data about individuals, said Maartje de Graaf, data protection lawyer at NOYB. “The technology must follow legal requirements, not the other way around.”
Innovation does not exempt us from laws
NOYB accuses OpenAI of providing false information about personal data in the case of an unnamed “public figure” without providing the legally required opportunity for correction or deletion. In the specific case, which also involved an incorrect date of birth, OpenAI argued that it was not possible to correct the data. You can block data for certain requests, such as the name of the celebrity. However, you cannot prevent ChatGPT from filtering all information about the complainant.
In NOYB’s opinion, OpenAI did not adequately respond to the complainant’s request for information. “It is of course possible to log the training data used to at least get an idea of the sources of information,” said de Graaf. “It seems that with every ‘innovation,’ a different group of companies feel their products don’t have to comply with the law.”
Activist successful against Facebook
Schrems had previously brought fear to the Facebook group Meta in two lawsuits and twice overturned important data agreements between the USA and Europe before the European Court of Justice. NOYB and the person concerned have now called on the Austrian data protection authority to investigate OpenAI’s data processing practices. Of particular interest is the question of what measures the start-up has taken to ensure the accuracy of personal data. A fine must be imposed on OpenAI to ensure future compliance with the regulations.
OpenAI caused a sensation in November 2022 with the introduction of ChatGPT: The program is able to use artificial intelligence to generate essays, poems or conversations from very short prompts.