“If it’s in our product, we need a patent.”
Lines like this come from decades of real-world patent strategy – strategy built around real-world technology like machines, hardware, and pharma. And they reflect assumptions about the risks and rewards of patenting in those industries.
But AI is different.
Technology is hidden, competitor activity is opaque, and products evolve rapidly. Patenting AI isn’t just “business as usual” – it’s a different game entirely.
Yet most AI patents I’ve seen still rely on outdated assumptions, which makes them less effective.
Here are six common assumptions that often don’t apply in AI—and what to consider instead:
1. “It’s in our product, so we need a patent.”
Traditional thinking: Exposure creates risk, so patent everything in the product.
AI reality: AI technology is often hidden behind APIs or models. Patenting exposes the invention and may create risk. Focus on patenting what will truly be visible or critical to competitors.
2. “Patents stop copying.”
Traditional thinking: A patent lets you stop anyone who copies your invention.
AI reality: Competitor technology is often invisible. You may never know if someone copies your work. Instead, patent only what can be detected if copied.
3. “Patents protect our market position.”
Traditional thinking: Protecting the product protects your edge.
AI reality: AI products evolve quickly. By the time a patent is granted, the product may have changed. Instead, patent the aspects that drive lasting customer value.
4. “Good patents block design-arounds.”
Traditional thinking: Clever claims will block all design alternatives.
AI reality: AI offers nearly limitless ways to design around a patent. Evaluate design-around risks honestly before filing, instead of assuming the patent will be the blocker you hope for.
5. “Breakthroughs need patents.”
Traditional thinking: Important technologies deserve strong patents.
AI reality: Value often lies in the use case, not the underlying code. Technology changes fast, but market fit proves real value. Patent the business opportunities unlocked by market-fit, and not just the technology itself.
6. “Strong patents are enforceable.”
Traditional thinking: The courts will protect your patent rights.
AI reality: Enforcement is inconsistent globally, and AI patents are vulnerable to invalidation in more ways than traditional patents. Patents are much more useful for business leverage than for litigation.
All patents may look the same on paper, but not all of them support your business – especially if you’re relying on old assumptions.
Effective AI patents are possible – essential even, but they require a reset back to first principles: Why are we patenting? What will actually achieve that? When is the right time?
Next time you’re considering filing, pause and rethink your assumptions.
The old rules don’t always apply anymore.