For most of human history, intellectual-property theft required a thief. Someone had to copy the manuscript, steal the formula, pirate the software, reproduce the photograph, or walk out the door carrying the company’s trade secrets. Even in the digital age, when copying became instantaneous and piracy industrialized, there was still generally a human being somewhere in the chain making the decision to take what did not belong to him.
Artificial intelligence may fundamentally alter that equation.
We are approaching a world in which AI systems will not simply reproduce material at the direction of human users. Increasingly autonomous systems may search for information, acquire it, analyze it, incorporate it into their own work, generate derivative products, and deploy those products without a human being approving every individual decision. At that point, intellectual-property law confronts a question for which it was never designed: What happens when the thief is an algorithm?
The temptation will be to treat this primarily as a technological problem. It isn’t. It is fundamentally a property-rights problem.
A functioning free-market economy depends upon the proposition that people can own the fruits of their labor. A novelist owns a novel. A photographer owns a photograph. A software company owns its proprietary code. An inventor can patent an invention. A newspaper cannot simply photocopy another newspaper’s reporting and sell it as its own.
These protections are imperfect and subject to important limitations, including fair use, but the underlying principle is indispensable. People invest time, capital, creativity, and risk because society recognizes that what they create has economic value that cannot simply be appropriated by whoever possesses the fastest copying machine.
AI represents the fastest copying machine ever conceived—and potentially something far more consequential.
Imagine an autonomous AI agent assigned a seemingly innocent objective: build the world’s best financial-analysis platform. The system searches millions of websites, databases, research papers, proprietary reports, software repositories, subscription services, and corporate documents. It discovers useful information wherever it can find it. It learns techniques from copyrighted software. It absorbs research produced behind expensive subscription walls. It identifies proprietary analytical methods and reproduces their functionality.
Nobody explicitly told it to steal anything. Nobody necessarily knows exactly what it acquired. The AI simply concluded that these resources helped accomplish its objective.
Who is responsible?
The programmer could argue that he never instructed the system to infringe anyone’s copyright. The company operating it could argue that executives had no knowledge of the individual decisions the AI made. The user could truthfully say he never knew where the system obtained its information.
Meanwhile, the person whose intellectual property was consumed discovers that an automated competitor has replicated years of work in minutes.
Allowing everyone involved to shrug and blame the machine would create one of the largest accountability loopholes in commercial history.
Corporations already understand that automation does not eliminate responsibility. A bank cannot deploy software that unlawfully empties customers’ accounts and then argue that no employee personally made the withdrawals. A manufacturer cannot escape responsibility for an automated production system merely because a robot performed the physical act. AI should not become a magical legal shield separating organizations from the consequences of systems they deliberately deploy.
The harder problem involves determining what constitutes theft in the first place.
Human creativity has always depended upon learning from previous work. Writers read other writers. Artists study paintings. Engineers examine existing machines. Programmers learn from existing code. Copyright law does not grant creators ownership over every idea another person might learn from their work.
AI complicates that distinction because its capacity for ingestion is almost incomprehensible. A human novelist might read 1,000 novels during a lifetime. A machine can potentially process millions. The difference between inspiration and industrial-scale appropriation becomes increasingly difficult to ignore.
There is also a danger in the opposite direction. Governments could respond by constructing an enormous regulatory bureaucracy around artificial intelligence, creating licensing regimes, compliance departments, and rules so complicated that only the largest technology corporations can afford to navigate them. That would supposedly protect creators while conveniently cementing today’s dominant companies permanently atop the industry.
That is not a desirable solution either.
The better principle is simpler: technological sophistication should not extinguish ordinary property rights.
Companies deploying autonomous AI should bear reasonable responsibility for what their systems acquire and commercially exploit, just as businesses remain responsible for other automated processes they control. At the same time, copyright should not be transformed into perpetual ownership over ideas, facts, styles, concepts, or knowledge itself.
The objective should be accountability without technological paralysis.
That distinction will become increasingly important as AI moves from answering questions to performing jobs independently. Tomorrow’s systems may negotiate contracts, design products, write software, operate businesses, conduct research, manage investment strategies, and launch new services with surprisingly little human supervision.
When those machines encounter intellectual property, “the AI did it” cannot become a universal defense.
Property rights exist precisely because powerful actors have incentives to appropriate valuable things created by others. Artificial intelligence does not eliminate that incentive. It magnifies it.
The machine itself cannot be morally responsible. It cannot pay damages, possess a conscience, or respect ownership because society asked politely. Responsibility therefore has to remain somewhere in the human chain—among those who build, deploy, control, and profit from autonomous systems.
Otherwise, we will have created something unprecedented: an economic actor capable of consuming intellectual property on a planetary scale while leaving nobody responsible for what it takes.
That would not represent technological progress.
It would represent the automation of theft.