Here is a new problem that companies around the world may face AI era.
A nefarious Big Tech employee leaves for a competitor with trade secrets and passes them on to a AI agent or model while employed by the competitor, and runs some tests using those secrets.
Perhaps the bad apple employee then uses that confidential knowledge to create a new solution, from which the competitor benefits. Or perhaps those secrets are stored in a knowledge base that an AI could retrieve when the nefarious employees’ colleagues have a relevant question.
Apple raised that possibility in a supplemental letter filed Monday in support of its request for expedited discovery in its filing trade secrets lawsuit against OpenAI.
In the filing, Apple’s lawyers said a former employee’s use of trade secrets while employed by OpenAI and his “use of AI agents to run simulations raise concerns beyond simple document theft.”
“When trade secrets are fed into an AI agent or model that ‘learns’ from them, such ‘learning’ can create irreversible and perpetually proliferating uses of the trade secret – harm that is, at the very least, uniquely challenging to undo and requires swift investigation,” Apple’s lawyers wrote.
The continued use of confidential information by a competing company is not a new problem. Artificial intelligenceHowever, introduces a new wrinkle into the matter: how should companies regain control of their secrets after entering a rival organization’s AI system?
New risk, same remedies
“Employees are already real loose cannons, walking around with knowledge in their heads,” says Camilla Hrdy, a law professor at Rutgers, whose work focuses on the law of trade secrets and generative AItold Business Insider. “Now they are putting that knowledge into AI, and that could mean a real loss of control. That is new.”
Elon Musk’s XAI raised a related, but distinctly AI-related concern when it sued OpenAI, accusing Sam Altman’s company of staff poaching steal Grok’s underlying technology.
That lawsuit stated that while Xuechen Li, a former xAI engineer, “had stored xAI’s entire codebase in his personal cloud storage account, Li also connected his personal ChatGPT account directly to his personal cloud storage account, set up as a connected ‘resource’ in ChatGPT from OpenAIIt added: “OpenAI had access to Li’s files, including the stolen copy of xAI’s entire source code, through its ChatGPT service.”
A judge dismissed the lawsuit in June.
Hrdy said these cases do not immediately call for new legal solutions. Possible solutions often include telling a company to stop using the trade secrets, not to reveal secrets, and to take steps to protect those secrets.
Damages must also be assessed, Hrdy said: actual losses suffered from the loss of those secrets, or, in some cases, royalties payable to the affected company.
Can an AI unlearn secrets?
Stopping a rival company’s use of trade secrets can pose technical challenges, depending on how exactly an employee applies confidential information to an AI system.
Sijia Liu, a professor of computer science at Michigan State University, co-authored a paper on “machine unlearning,” the process of removing the influence of certain data or capabilities from an AI model.
He told Business Insider that if a document containing sensitive information is stored in a repository that an AI system can pull from, the solution could be as simple as deleting the file.
On the other hand, if sensitive information were to be used to train or refine a model, it would require a completely different, and likely labor-intensive, process.
“The second case could be more difficult because the influence of something is very difficult to evaluate,” Liu said, adding that “you need to precisely define the boundary of undesirable possibilities.”
A more direct approach to hiding secrets could be to build a “detection system” that flags sensitive user requests or sensitive information passed between agents, Liu said. The detector can then activate a hard stop in response to the request. Liu said this is not “true unlearning,” but it is more practical.
To be clear, Apple hasn’t said how the former employee might have used trade secrets with an AI, whether it was a one-off AI-assisted simulation or if there was training that could impact a broader model.
An Apple spokesperson did not return a request for comment on this story.
Either way, AI can reveal new ways for companies to lose control of their secrets.
