A recent US federal court case involved a former marketing agency employee who used stolen trade secrets to train an AI agent, aiming to replicate his employer’s expertise and compete directly. The court granted a temporary restraining order and preliminary injunction, calling the AI agent an existential threat to the business. The defence’s claim that no harm occurred because files were placed in a “quarantine folder” was rejected.
The case shows that AI can be used to misappropriate trade secrets in ways that are difficult to detect or reverse. Even if stolen files are removed, the AI model may still be influenced by the data, making it hard for defendants to prove otherwise. The risk extends to inadvertent violations, as employees using personal devices with AI agents could expose confidential information without realising it.
For regulated companies, standard digital discovery is no longer enough. Full forensic imaging and rapid legal action are now minimum requirements. AI also makes it easier to reverse-engineer products, increasing uncertainty in IP litigation. Firms should act immediately if data misuse is suspected and update security and legal policies to address these new risks.
