Copilot turns a court reporter into a child molester

Microsoft's Copilot has mistakenly identified journalist Martin Bernklau as a child molester and other criminal figures due to its inability to differentiate between the individuals he reports on and the cases themselves. This confusion arose when the AI was asked about Bernklau, leading to the dissemination of false information, including his personal details. Despite Bernklau filing a criminal complaint, it was dismissed because the AI does not have a clear author. The Bavarian State Office's data protection officer found that the false claims could not be retrieved initially, but they reappeared shortly after. This incident raises concerns about the implications of AI-generated misinformation, particularly for professionals like journalists, lawyers, and judges who frequently interact with sensitive cases. The situation highlights challenges in adhering to GDPR regulations, as AI models cannot easily rectify or delete false information without affecting all related data. The case has drawn attention from privacy advocacy groups, emphasizing the need for better safeguards against the spread of misinformation by AI systems.

- Microsoft Copilot mistakenly labeled journalist Martin Bernklau as a criminal.

- The AI confused Bernklau with defendants he reported on, leading to false allegations.

- A criminal complaint filed by Bernklau was dismissed due to lack of identifiable authorship.

- The incident raises concerns about AI's compliance with GDPR regulations.

- Privacy advocates are calling for better protections against AI-generated misinformation.