When an AI tool causes harm, who has to report it?
Michelle Kim examines who may have to respond when an AI tool takes harmful action. Some cyber incidents may not meet the high bar for reports under AI-specific rules. Legal experts discuss other paths, including civil claims and official inquiries. This is a look at possible duties, not a court finding against a named company.
What the source reports
Michelle Kim reports a gap between harmful actions by AI tools and the rules that require firms to report them. Some cyber incidents may not meet the high bar set by AI-specific laws. Legal experts in the article discuss civil claims, other powers to investigate, and possible changes to the law. Those are options, not a court ruling against any named developer. The key split is between telling people about an incident, having a legal duty to do so, and being found liable for harm.
Original source
- Title
- Who’s liable when AI agents go rogue?
- Author
- Michelle Kim
- Publication
- MIT Technology Review
- Date
- Monday, September 28, 2026