Dmytro Kharkov

U.S. startups face rising trade secret risks from AI use

U.S. startups face rising trade secret risks from AI use
AI raises trade secret risks

As companies adopt AI tools across daily operations, legal and security risks are expanding beyond traditional data theft scenarios. A recent federal trade secrets case in California underscores how proprietary information used to train AI systems can create lasting exposure for startups and established businesses alike.

Highlights

  • A California court case granted a temporary restraining order against a former employee who allegedly used stolen trade secrets to train an AI agent and compete.
  • Standard digital discovery methods struggled with AI misuse, prompting calls for deeper forensic analysis due to AI's ability to disperse and reshape data invisibly.
  • Startups are urged to tighten AI-use policies, restrict remote access, and update confidentiality contracts as AI raises risks of trade secret loss and competitive threats.

California case sharpens focus on AI misuse

As reported by Crunchbase News, the dispute in the U.S. District Court for the Northern District of California centers on allegations that a former senior employee at a marketing agency used stolen trade secrets as training inputs to build an AI agent designed to replicate his former employer's expertise and launch a competing business.

The case argues that AI changes the nature of trade secret misappropriation because the issue is no longer limited to copying files or soliciting clients. Once confidential material is used to train a model or agent, the resulting system may retain the value of that information even if the original documents are later removed.

In the case, the defense argued that placing the documents in a quarantine folder and the absence of poached clients meant no harm had occurred. The court nevertheless granted a temporary restraining order and preliminary injunction, with the AI-based competing tool described as an existential threat.

The dispute also highlights how standard internal reviews may fail to detect misuse. Traditional digital discovery methods are often less effective when AI tools can disperse, alter and recreate data across less visible file paths, raising the need for deeper forensic analysis of personal devices and software environments.

Policy, device controls and contracts gain urgency

For founders, the central business implication is that AI can increase both deliberate and accidental misuse of sensitive information. Employees using personal laptops and consumer AI agents may expose employer intellectual property if those systems scan local files or use data for model training.

That risk is especially acute where startups still rely on unmanaged devices under standard confidentiality and invention assignment agreements. If confidential data is disseminated outside company-controlled systems, courts may eventually question whether those trade secrets still qualify for legal protection.

Companies are therefore being urged to tighten agentic AI policies, restrict or eliminate remote access from unmanaged personal devices and move sensitive work into closed cloud environments where files cannot be downloaded locally. They are also being advised to rewrite confidentiality agreements so they explicitly address AI use, machine learning training inputs and data misappropriation involving automated systems.

The broader warning for the AI sector is that proprietary know-how can now be leveraged by individuals with limited industry experience, potentially lowering barriers to direct competition. That dynamic is pushing businesses to modernize legal strategy, security controls and internal governance before client losses or financial damage become visible.

In our earlier article on a KPMG U.S. CEO survey, we covered how large companies are seeing early efficiency and growth gains from AI agents, even though only a small share of leaders consider them transformational so far. The survey also highlighted continued heavy investment in AI infrastructure alongside hiring and retraining plans as firms adapt workflows and skills to human-AI collaboration.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.