Can AI Be an Inventor? DABUS Case Explained
“Understand the DABUS case, global patent rulings, and how to legally protect AI assisted inventions”
Introduction: The day AI put your patent at risk
Imagine this, You stay up late at your desk, your screen glowing with code, prompts, and 3D‑rendered prototypes. Your AI model is humming along, and suddenly it suggests a design you’ve never seen before—sleek, functional, and different from anything on the market. Your heart jumps.
You rush to your team, show them the output, and the room lights up. Someone jokes, “Did the AI just invent this?”
The room laughs.
Then someone asks, quieter:
“…So who actually invents this on the patent?”
That joke is not a joke.
It’s a question that has already played out in courtrooms and patent offices around the world.
Today, AI is no longer a silent assistant in the background.
It is in the lab, in the design studio, and in the R&D pipeline—often doing the kind of work that used to belong only to human inventors. Yet when it comes to the law, there is a hard rule: only a human can be an inventor.
The landmark DABUS case—where an AI system was named as the sole inventor on patent applications—made this rule visible to everyone. Courts and offices in the US, Europe, the UK, and other major markets all sent the same message:
AI can invent, but it cannot be the inventor.
For innovators, this creates a quiet tension:
You want to push the limits of what AI can do.
But you also need to protect your life’s work with valid, enforceable patents.
In this article, you will learn:
- Why AI cannot be an inventor under current law,
- What the DABUS case actually means for your startup,
- And how to protect your AI‑assisted inventions without risking your patent or your ownership.
Because the future of invention is not human or machine.
It’s human with machine.
And your IP strategy must reflect that truth—before you hit “file” on your next patent application.
Table of Contents
How can we help you?
Contact us for free consulation
Why this matters now
AI is no longer just a simple tool.
Startups, companies, and research teams use AI to:
- Generate new product shapes and structures
- Optimize mechanical, electrical, or software designs
- Find technical solutions humans might miss
But under most patent laws today, only a human can be an inventor.
If you get inventorship wrong, the patent can be rejected, challenged, or even cancelled later.
The risk is real:
- No patent protection for your core innovation
- Ownership disputes between founders, employees, or investors
- Loss of competitive advantage in fast‑moving markets
You can use AI to invent faster.
But the law still says: inventorship belongs to humans.
The DABUS case: The AI that tried to be an inventor
To test this question, one case became the global example: DABUS.
Dr. Stephen Thaler, an AI researcher, created DABUS—a system designed to generate inventive ideas on its own.
It produced at least two outputs that looked like clear inventions:
- A fractal‑based food container with better grip and improved heat performance
- A flash‑activating light device that could help in emergencies
Dr. Thaler filed patent applications in many countries and did something unusual:
He listed DABUS, the AI, as the sole inventor.
This small move started a global legal debate.
What patent offices around the world decided
Countries and offices asked the same key question:
Can a non‑human system, such as an AI, be an “inventor” under patent law?
The answers were very similar:
- United States (USPTO + Federal Circuit)
Courts and the USPTO said an inventor must be a natural person. AI cannot be an inventor. - European Patent Office (EPO)
The EPO refused the DABUS‑related applications, ruling that inventors must have legal capacity, which AI systems do not have. - United Kingdom (UKIPO + Supreme Court)
The UKIPO and later the UK Supreme Court held that only a human can be an inventor and assign rights. - Australia
A lower court first allowed AI as inventor, but the decision was overturned on appeal. The final view is the same as other major systems: inventors must be human. - South Africa
One DABUS‑linked patent was granted naming the AI as inventor, but this was mostly a procedural result in a system with limited examination, not a legal recognition of AI inventorship.
The global message is clear:
Under current patent law, AI cannot be an inventor.
Why the law says “no” to AI inventors
The rejections are based on basic legal ideas:
- No legal identity
- AI systems are not legal persons.
- They cannot own property, sign contracts, or be treated as rights‑holders.
- Ownership and assignment become unclear
- Patents start from the inventor, who then usually assigns them to a company or employer.
- If the “inventor” is not a recognized legal subject, ownership and assignment are weak or invalid.
- No way to hold AI accountable
- Inventors can be asked to explain their work, prior‑art knowledge, and experiments.
- AI systems cannot be called on to answer, defend, or correct their “inventions.”
In short:
AI can help create ideas, but it cannot act as a legal person in the patent system.
Where AI fits in a patent application
Even if AI cannot be the inventor, AI‑assisted inventions can still be patented.
The rule is simple:
- A human must be the inventor (the one who conceived the invention).
- AI can be the tool that helps find, test, or refine the idea.
In practice:
- If a designer or engineer uses AI to create a new structure, chooses the best outputs, adjusts parameters, and completes the solution, that engineer is the inventor.
The AI is, legally, the same kind of tool as CAD software, simulation code, or advanced lab equipment.
How innovators should protect AI assisted inventions
If you use AI to invent, follow a clear, simple plan.
- Identify the human inventor(s)
- Ask: Who set the problem, designed the AI workflow, and chose the final invention?
- That person is the inventor, even if the AI did much of the computation.
- Keep clear records
- Save notes, emails, and logs showing:
- Human decisions on goals and constraints
- How AI outputs were filtered and improved
- This helps support your inventorship claim if challenged later.
- Save notes, emails, and logs showing:
- Treat AI as a tool, not a co‑inventor
- In internal discussions and training, describe AI as:
- A powerful assistant,
- A design or testing tool,
- Not a legal person.
- This keeps your team aligned with the law.
- In internal discussions and training, describe AI as:
- Draft patent applications with human‑centric language
- Claims should focus on what the human did:
- “A method in which a designer uses an AI model to… and then selects…”
- You can describe the AI in background or implementation sections, but not as the source of inventive thinking.
- Claims should focus on what the human did:
- Plan for different countries
- Most major systems agree today, but laws can change.
Use tools like PCT or Paris routes, follow local AI‑related updates, and consider other IP paths (such as trade secrets or design protection) where useful.
What happens if inventorship is wrong
Mistakes in inventorship can cause serious problems:
- Application rejection due to inventorship errors
- Ownership disputes between founders, employees, or investors
- Delays in product launch while disputes are resolved
- Loss of market edge if a patent is later cancelled
In the DABUS‑related cases, naming the AI as inventor led to broad rejection across many countries—a clear warning for anyone using AI‑driven R&D.
How Novel Patent Services Helps
If you are working on an AI-assisted invention, identifying inventorship correctly and drafting the application properly is critical.
Novel Patent Services supports inventors, startups, and businesses with patent drafting and filing support for AI-assisted inventions, helping you reduce risk and prepare stronger applications aligned with current legal requirements.
The future of AI and patent law
The DABUS case did more than answer one question.
It started a global discussion:
- Should the law change to allow AI‑related inventorship?
- Should AI‑only outputs get new types of IP rights, separate from patents?
- How should patent office’s handle AI‑assisted versus AI‑only inventions?
Changes may come in the future.
But today the rule is clear:
AI can help invent—but it cannot be the inventor under current patent law.
A simple takeaway for innovators
To put it plainly:
- AI is changing how inventions are created.
- Law is still catching up—but it is clear today.
- Inventorship is human‑centric.
For you, as an innovator or startup founder, the steps are:
- Use AI strongly to speed up design and discovery.
- Secure your IP through human inventors and clear records.
- Match your story to the law—so your next breakthrough is protected, not at risk.
Because the future of invention is not human or machine.
It’s human with machine.
And your IP strategy must reflect that truth—before you hit “file” on your next patent application.
Protect Your AI-Assisted Invention Before You File
Using AI does not stop you from getting a patent. But getting inventorship wrong can create serious risks.
If you want to move forward with clarity, Novel Patent Services can support you with patent drafting and filing for AI-assisted inventions.
Get in touch with us
FAQ'S
No. Under current patent laws in the US, Europe, UK, and many other countries, only a natural person can be an inventor. AI systems cannot be named as inventors.
- Ownership usually belongs to the human inventor or to their employer, based on contracts or assignment agreements. The AI itself cannot own rights.
Yes, if they meet standard requirements such as novelty, inventive step (non‑obviousness), and industrial applicability, and if the inventor is a human.
The DABUS case involves an AI system created by Dr. Stephen Thaler that was named as an inventor in patent applications. Major patent offices and courts rejected this, confirming that inventors must be humans under current law.