The ChatGPT language model is able to generate coherent and convincing texts, but its use raises ethical and legal questions. It might be banned in Europe due to concerns about the collection and use of personal data.
OpenAI faces investigations in several European Union countries due to concerns about the collection and use of personal data by its text-generating chatbot, ChatGPT.
The Italian authority has temporarily blocked the use of ChatGPT, while data regulators, German, Irish and Canadian are also investigating how OpenAI collects and uses data. The European Data Protection Board, the umbrella organization for data protection authorities, has also set up a working group to coordinate investigations and enforcement around ChatGPT, while France’s CNIL has opened an investigation.
Will ChatGPT be banned in Europe? OpenAI has 10 days to comply with EU requirements
OpenAI’s thirst for larger models, based on massive data collection from the internet, is the root of the company’s current problems. Experts believe that OpenAI Compliance to European data protection rules will be practically impossible. Regulators are asking OpenAI to seek people’s consent for the use of their data, to inform them about the use of their data, to give them the power to correct errors and to delete their data if they wish .
And OpenAI fails to prove that its data use practices are lawful, it risks facing fines, having to delete data, or even being banned in certain countries or throughout the European Union. OpenAI’s alleged violations are so egregious that they could end up in the Court of Justice of the European Union.
To read: ChatGPT: 4 things not to do when using the famous AI
The EU General Data Protection Regulation is the strictest data protection regime in the world and its scope has been widely copied around the world. Data regulators in many countries will be watching the outcome of this case, and the outcome could change the way AI companies collect data.
OpenAI will have to be more transparent about its data practices and will need to show that it uses one of two possible legal ways to collect training data for its algorithms: consent or legitimate interest. OpenAI is unlikely to be able to prove that it obtained data subjects’ consent for the collection of their data, which would leave the company with the argument that it had a legitimate interest in doing so. This would require the company to make a compelling case to regulators about the real importance of ChatGPT to justify collecting data without consent. Italy has given OpenAI until April 30 to comply with the law.
A needle in a haystack
Tech companies tend to not documenting how they collect or annotate data for machine learning. They often don’t even know what’s in the dataset, according to Nithya Sambasivan, a former Google researcher and entrepreneur who studied AI data practices, speaking to the MIT Technology Review.
Find the data of Italians in the extensive training data set of ChatGPT is like looking for a needle in a haystack. Even if OpenAI managed to delete user data, it is not certain that this step is permanent. Studies have shown that data sets persisted on the Internet long after they were deletedas copies of the original tend to remain online.
To read: ChatGPT, Bard: for the boss of Google, AI requires the same regulations as nuclear weapons
« Data collection is very, very immature when it comes to AI, says Margaret Mitchell, AI researcher and chief ethics scientist at startup Hugging Face, who was previously co-head of AI ethics at Google. Indeed, while tons of work has gone into developing state-of-the-art techniques for AI models, data collection methods have not changed much over the past decade.
« What’s really concerning is how he uses the data you give him in the chat », Explains Alexis Leautier, AI expert at the CNIL. People tend to share intimate and private information with the chatbot, telling it about things like their mental state, health, or personal opinions. Leautier says it’s problematic if there’s a risk that ChatGPT will regurgitate this sensitive data to others. And under EU law, users must be able to have their chat log data deleted, he adds.
