I believe this will become one of the defining governance challenges businesses face over the next decade.

Traditionally, intellectual property misappropriation was largely associated with copying an identifiable asset: a confidential document, source code repository, product design, business strategy, database, or customer file.

Investigations focused on locating the copied artefact.

AI changes that model.

Today, a person with access to valuable company information can, intentionally or otherwise, provide that information to an AI assistant and ask it to:

• Explain it
• Summarise it
• Restructure it
• Rewrite it
• Generate an alternative implementation
• Recreate its functionality
• Remove obvious references to its original source

Consider a proprietary software platform.

Instead of simply copying source code, someone could allow an AI coding agent to analyse the architecture, business rules, APIs, implementation patterns and design decisions, and then generate a different-looking implementation that performs substantially the same functions.

No identical copy of the original repository may later be found.

The filenames, code structure, wording and hashes may all differ.

Yet much of the underlying knowledge, engineering effort, design thinking and commercial advantage may still have been exposed, extracted and reconstructed.

The same principle extends far beyond software.

A strategic planning document can become a new market-entry strategy.

• A pricing model can become a competitor’s pricing framework.
• A confidential operating procedure can become a rewritten process manual.
• A proprietary methodology can become a new consulting framework.

The wording changes.

The filenames change.

The underlying intellectual capital may not.

I use the term IP distillation to describe the extraction and reconstruction of the knowledge, reasoning, methods, trade secrets and commercial value embodied in intellectual property through AI, without necessarily reproducing the original artefact itself.

Recent work in AI governance and digital forensics has reinforced for me just how significant this emerging risk may become.

We are examining how organisations can identify AI-assisted transformations of sensitive information, investigate them responsibly, and strengthen the governance, provenance, and auditability of enterprise AI systems.

This is not an argument against AI.

AI is one of the most important technologies of our generation. It will transform how organisations innovate, build products and create value.

The challenge is ensuring that organisations have the governance, auditability and visibility required to use AI responsibly while protecting the intellectual property that gives them their competitive advantage.

Traditional security controls answered the question:

“Was the file copied?”

The AI era requires us to answer a different question:

“Was the knowledge extracted?”

I believe the next generation of enterprise AI governance will need to answer questions such as:

• What company information was provided to an AI system?
• Which documents, repositories or workspaces were used as context?
• How was that information transformed?
• What outputs were generated?
• Were those actions authorised and consistent with company policy?
• Can the organisation demonstrate provenance?
• Can it reconstruct the complete chain from source asset to AI context, user prompt, generated output and final implementation?

Protecting intellectual property in the AI era will require more than access control.

It will require traceability:

Source asset → AI context → User prompt → Generated output → Human acceptance → Repository commit → Production deployment

I suspect this will become one of the defining enterprise-governance challenges of the AI era.

I’m interested to hear whether others are seeing similar concerns emerge within their organisations, and how they are thinking about protecting intellectual property in an AI-first world.