Copilot Falsely Accuses Journalist Who Is Now Suing Microsoft

AI-generated defamation incident: A German journalist, Martin Bernklau, became the victim of false and defamatory statements generated by Microsoft’s Copilot AI, raising concerns about the responsibility of AI companies for the content their systems produce.

• Bernklau, who has decades of experience reporting on criminal trials, discovered that Copilot AI had falsely accused him of various crimes, including child abuse and exploiting widows as an undertaker.
• The AI system mistakenly attributed crimes Bernklau had reported on to the journalist himself, conflating the reporter with the subjects of his articles.
• In addition to the false accusations, Copilot also disclosed personal information of the journalist, including his phone number and address, further compromising his privacy.

Legal and regulatory response: The incident has prompted reactions from legal authorities and data protection agencies, highlighting the complexities of addressing AI-generated misinformation.

• The Tübingen Public Prosecutor’s Office stated that no criminal offense had been committed since no actual person had made the allegations, illustrating the legal challenges in dealing with AI-generated content.
• The Bavarian Data Protection Office intervened by contacting Microsoft, which initially filtered Copilot’s replies about Bernklau, but the false information reappeared days later.
• Bernklau has taken legal action, hiring a lawyer and suing Microsoft for defamation and invasion of privacy, challenging the company’s attempt to disclaim responsibility through its terms of service.

AI system responses and ongoing issues: The incident has led to evolving responses from AI systems and raised questions about the persistence of false information.

• Following media attention, Copilot now acknowledges the controversy, stating that Bernklau was falsely labeled by an AI chat system as a convicted child molester, psychiatric facility escapee, and fraudster.
• Perplexity, another AI search system, returns similar results but explicitly names Copilot as the responsible chat system, indicating the spread of information about the incident across AI platforms.
• The reappearance of false information after initial filtering attempts suggests ongoing challenges in consistently preventing AI systems from generating and disseminating misinformation.

Broader implications for AI responsibility: This case underscores the critical need for clear accountability frameworks and robust safeguards in AI systems, especially those interacting directly with the public.

• The incident highlights the potential for AI systems to cause significant harm to individuals’ reputations and privacy, raising questions about the ethical and legal responsibilities of AI companies.
• It also exposes the limitations of current legal and regulatory frameworks in addressing harm caused by AI-generated content, potentially spurring discussions about new laws or regulations.
• The case may serve as a precedent for how similar incidents are handled in the future, potentially influencing the development of AI governance and liability policies worldwide.