OpenAI is facing a privacy complaint in Europe after its AI chatbot ChatGPT hallucinated a false murder allegation.

A Norwegian individual is being backed by data privacy advocacy group Noyb after ChatGPT generated a false response that the man was convicted of killing his two children, and attempting to murder his third.

This is one of ChatGPT’s more disturbing hallucinations – which is the technical term for when generative AI’s generative musings go completely off the rails, spurting out complete falsehoods.

Previously, the chatbot has provided incorrect birthdates and other biographical facts, but this hallucination said that an innocent man had murdered his own children, and was convicted of a crime that was not even committed.

The query that prompted the shocking response was quite simple as well – the individual simply asked ChatGPT who he was, which immediately prompted the chatbot to respond with a full story about a murder case that never happened.

In the past, OpenAI has offered to block responses that show incorrect information, but does not allow individuals to correct incorrect information its chatbots generate.

This is against the EU’s GDPR, which provides individuals the right to rectify personal data mistakes.

GDPR also stipulates that data controllers – which would include OpenAI – ensure that the personal data they have or produce about an individual is correct.

The Noyb case filing posits that OpenAI is therefore in breach of two key GDPR stipulations about personal data accuracy.

“The GDPR is clear. Personal data has to be accurate,” Joakim Söderberg, data protection lawyer at Noyb, said in a statement.

“If it’s not, users have the right to have it changed to reflect the truth. Showing ChatGPT users a tiny disclaimer that the chatbot can make mistakes clearly isn’t enough. You can’t just spread false information and in the end add a small disclaimer saying that everything you said may just not be true.”

While a breach of GDPR can lead to a fine of up to 4% of a company’s annual global turnover, a lawsuit of this severity could bring into question the legality of OpenAI’s model when it comes to accuracy assurances and personal information production.

Recommended

• EU to Change GDPR Cross-Border Regulations
• OpenAI Faces Lawsuit Over Potential ChatGPT GDPR Violations
• EU Creates ChatGPT Task Force to Keep Up with AI Developments

OpenAI has already faced a range of data privacy roadblocks on its way to AI domination. It faced a temporary block in Italy, and then a fine of 15 million euros for using people’s personal data without a proper legal basis under GDPR.

The EU is known to set the standard for global regulation, taking a relatively cautious approach to emerging technology’s roll out, and having some of the most strict data protection regulations for companies in the world.

Still, the EU appears to be grappling with just how to regulate generative AI, with its unprecedented proliferation and potential for economic growth, its developing societal disruption, and budding penchant for chaos – be it in the global market, the surge of deepfakes, or its use by cyber-criminals.

The EU seems stumped in other GDPR cases having to do with the world’s most popular AI chatbot. Poland’s data protection board has yet to decide on a privacy complaint made against ChatGPT back in 2023, and the Irish data watchdog has urged against a headstrong approach to regulating generative AI.

The new lawsuit, brought up by data privacy titans Noyb, could be a watershed moment for the EU to finalise its position on just how much generative AI can get away with before GDPR must put its foot down.