A continuaciónSiguiente guía
Abogados sancionados por citaciones de casos fabricados con IA
sociedad
GUÍA de sociedad
IA e invención de patentes: los casos DABUS
No, an AI system cannot be listed as an inventor on a patent.
Escuche esta guíaNarrado por IA · ~3 min
En esta pagina4 minutos de lectura
Descripción general
Courts and patent offices in the United States, the United Kingdom, the European Patent Office, Germany and Australia have all ruled that an inventor must be a natural person, rejecting Stephen Thaler's applications that named his AI system DABUS. Inventions made with AI can still be patented when a human makes a significant contribution, which matters to every company that uses AI in research and development.
Buceo profundo
DABUS, short for Device for the Autonomous Bootstrapping of Unified Sentience, is an AI system built by the American computer scientist Stephen Thaler. In 2018 and 2019 Thaler filed patent applications in many countries for two inventions he said DABUS came up with on its own: a food container with a fractal-shaped surface, and a flashing light designed to attract attention in emergencies. He named DABUS as the sole inventor and himself as the owner. He meant these filings as test cases. The results were almost all the same. The European Patent Office refused the applications, and in 2021 its Legal Board of Appeal confirmed that an inventor under the European Patent Convention must be a human being. In the US, the Federal Circuit held in Thaler v. Vidal (2022) that the Patent Act's word "individuals" means natural persons, and the Supreme Court declined to hear the case in 2023. In December 2023 the UK Supreme Court ruled unanimously that an inventor under the Patents Act 1977 must be a natural person. In Australia a single judge accepted DABUS as an inventor in 2021, but the Full Federal Court reversed that decision in 2022. Germany's Federal Court of Justice (2024) also rejected an AI inventor, while indicating that a human could be named even where AI was used. South Africa granted a patent in 2021, but it carries little weight because South Africa does not examine the substance of patent applications. Many people assume these rulings make inventions produced with AI help unpatentable. They do not. The courts decided who can be named as an inventor, not whether using AI disqualifies an invention. What matters now is whether a human contributed enough to count as an inventor. A second misconception is that DABUS lost because the inventions were weak. The cases turned almost entirely on the legal wording that defines who an inventor is.
Implementación en el mundo real
A pharmaceutical team uses a generative model to propose thousands of candidate molecules. A chemist then picks one and changes it based on her own hypothesis about how it binds. Her contribution can support her being named as inventor, but naming the model would get the application rejected.
An engineer types "design a better bicycle brake" into an AI tool and files whatever comes back. He risks having no qualifying human inventor, because simply presenting a general problem is unlikely to count as a significant contribution.
A patent attorney preparing inventor declarations for an AI-heavy project records who wrote the specific prompts, who judged and refined the outputs, and who built and tested the working prototype.
A startup filing in the UK and at the European Patent Office names its lead researcher, not its AI platform, as inventor. The UK Supreme Court's 2023 ruling and the EPO's decisions made clear that an AI cannot be designated.
The Future of AI and Patent Inventorship: The DABUS Cases
In the major jurisdictions the rule that inventors must be human now looks settled. Further change is more likely to come from legislatures and patent offices than from new court cases. Several questions remain open. How much human input is enough as AI systems take on more of the design work? Will widespread AI use change the "person skilled in the art" standard used to judge whether an invention is obvious? Will some inventions made with little human input fall outside patent protection and be kept as trade secrets instead? Some commentators have proposed a new type of protection for AI-generated inventions, but no major jurisdiction has adopted one. Patent office guidance is likely to keep changing as examiners see more AI-assisted applications.
Poniéndolo en práctica
Por qué es importante
- Riesgo y seguridad. Los daños catastróficos y cotidianos de la IA dependen de quién comprende los riesgos y quién puede actuar.
- Decisiones más claras. La alfabetización pública y profesional determina si es políticamente posible una política de seguridad sólida.
- Cortando el bombo. Las explicaciones claras reducen la captación por la exageración, las relaciones públicas de laboratorio y el vago teatro de ética.
cuidado con
- Tratar el riesgo existencial como ciencia ficción mientras que la capacidad se agrava.
- Confundir la seguridad del producto superficial con la alineación en condiciones de alta autonomía.
- Dejando a las audiencias que no hablan inglés ni a expertos solo con fuentes de baja calidad.
Próximos pasos
- Separe los riesgos de daños al producto, mal uso y pérdida de control/desalineación.
- Pregunte qué evidencia cambiaría su opinión sobre los plazos y la gravedad.
- Prefiera fuentes primarias y evaluaciones concretas a afirmaciones de marketing.
- Identifique un camino de acción: carrera, política, financiamiento o habilidades, no solo concientización.
Sigue explorando
Boletín gratuito
Recibe el resumen diario de IA
Tres noticias verificadas sobre IA cada mañana de lunes a viernes, explicadas con claridad. Gratis para siempre y sin anuncios.
Un correo cada día laborable. Date de baja con un clic. Nunca vendemos ni compartimos tu dirección.
Ponte a prueba
Take the AI and Patent Inventorship: The DABUS Cases quiz
Corrección inmediata en cada respuesta y un certificado para compartir, con un ID verificable, cuando apruebas un curso.
Apoya la educación gratuita sobre IA. AI Understanding es una organización sin fines de lucro 501(c)(3): sin anuncios ni muros de pago, nunca. Hacer una donación
Preguntas frecuentes
What is AI and Patent Inventorship: The DABUS Cases?
No, an AI system cannot be listed as an inventor on a patent. Courts and patent offices in the United States, the United Kingdom, the European Patent Office, Germany and Australia have all ruled that an inventor must be a natural person, rejecting Stephen Thaler's applications that named his AI system DABUS. Inventions made with AI can still be patented when a human makes a significant contribution, which matters to every company that uses AI in research and development.
¿Qué dos inventos dijo Stephen Thaler que DABUS creó por sí solo?
Las aplicaciones de Thaler abarcaban un recipiente de comida con forma fractal y una luz intermitente para llamar la atención en caso de emergencia. Los inventos en sí fueron modestos. Los casos trataban sobre quién podría ser nombrado inventor.
¿Qué decidió el Circuito Federal de Estados Unidos en Thaler v. Vidal (2022)?
El tribunal interpretó que la palabra legal "individuos" significa seres humanos, por lo que una IA no puede ser un inventor. La Corte Suprema se negó a revisar la decisión en 2023.
¿Qué dictaminó la Corte Suprema del Reino Unido en diciembre de 2023?
La Corte Suprema del Reino Unido dictaminó por unanimidad que la Ley de Patentes de 1977 exige que el inventor sea una persona física, lo que coincide con los resultados en los EE. UU. y en la OEP.
¿Por qué la patente DABUS de Sudáfrica de 2021 tiene poco peso legal?
Debido a que Sudáfrica otorga patentes sin examinar su sustancia, la concesión no fue una decisión considerada de que la IA puede ser inventora.
¿Cómo fue el litigio DABUS en Australia?
Australia pareció brevemente una excepción después de que un juez de primera instancia aceptara DABUS. El Pleno del Tribunal Federal anuló esa decisión en 2022.
sigue aprendiendo
Guías relacionadas
Más guías seleccionadas para este tema.