Patents For AI Inventions: A Comparison Of Requirements In Brazil, USA, And Europe

Brazil and Europe emphasize technical contributions with an industrial or reproducible application, while the USPTO requires a demonstrable technical improvement in an applied field

Innovation in artificial intelligence (AI) is reshaping industries, leading jurisdictions like Brazil, the US, and Europe to establish guidelines for patenting AI-related inventions. Despite shared objectives, their approaches differ regarding patentability, inventiveness, and disclosure requirements.

Brazil: Technical Effect and Software Limitations

In Brazil, AI-related inventions fall under the Industrial Property Law (Law No. 9,279/1996) and BPTO guidelines, notably the Guidelines for Examining Patent Applications Involving Computer-Implemented Inventions (2020). While software “per se” is excluded, AI systems that produce a technical effect beyond mere data processing may be patentable.

BPTO guidelines state that AI techniques used to solve technical problems can be considered inventions. Therefore, patent applications must clearly describe AI’s technical and innovative role, demonstrating a tangible industrial application. Recent appellate BPTO decisions emphasize the necessity of a technical effect, requiring AI inventions to be implemented in a non-obvious and innovative manner.

The importance of clear and precise technical descriptions in AI patent applications cannot be overstated. Patent applicants must detail how their AI solutions interact with hardware or other systems to produce the claimed technical effect. Furthermore, they should highlight specific improvements over prior technology, illustrating how AI optimizes processes, enhances efficiency, or provides innovative functionalities in industrial applications.

Currently, Brazil lacks specific AI-related patent regulations, but the BPTO is expected to introduce new guidelines by 2025, potentially refining eligibility criteria. Stakeholders are closely monitoring these developments to understand whether stricter or more flexible standards will be adopted, shaping the future landscape of AI patentability in Brazil.

United States: USPTO Eligibility and Practical Applications

The USPTO assesses AI-related patents based on subject matter eligibility and practical application. The 2024 updates reinforce a two-prong approach:

1. Prong One: Determines if a claim includes a judicial exception (e.g., an abstract idea or mathematical concept).

2. Prong Two: Evaluates whether the exception is integrated into a practical application with a tang ible technical effect, such as improvements in speech recognition or network security.

This framework, derived from the Alice/Mayo test, ensures that AI-related patents offer concrete technical advancements rather than claiming abstract ideas. AI inventions must solve specific technical problems and provide functional improvements in established technical fields.

To increase the chances of obtaining a patent in the US, applicants should ensure that their claims emphasize technical applications rather than broad, conceptual AI functionalities. The USPTO places particular weight on whether AI contributes to measurable improvements in efficiency, accuracy, or system functionality. AI innovations related to cybersecurity, autonomous systems, or medical diagnostics, for example, often demonstrate stronger patentability due to their direct technical benefits.

USPTO examiners also assess how AI interacts with underlying hardware or networks. Inventions that optimize computational performance, enhance data security, or improve user interactions through technical implementations are more likely to be considered patentable. The ability to demonstrate a real-world, practical advantage of AI over existing methods remains a key determinant in securing patent protection in the US.

Europe: Technical Purpose and Detailed Disclosure

The European Patent Office (EPO) requires AI inventions to serve a “specific technical purpose,” meaning they must contribute to a defined technical solution, such as advancements in image or audio processing. The EPO guidelines highlight two key factors:

• Technical Character: AI applications must go beyond standard algorithms, integrating into a practical context.
• Reproducibility: The invention must be disclosed in sufficient detail, including algorithmic structures and data characteristics, ensuring that a skilled person can replicate it.

Unlike in the US, where data specifics are not mandatory, the EPO requires detailed descriptions of training data characteristics (e.g., type, quality, and distribution), ensuring AI models can be reliably reproduced.

A major emphasis in European patent applications is ensuring transparency in AI implementation. Applicants must describe not only the functionality of the AI but also the way the AI interacts with other components to produce a specific effect. For example, an AI system that improves medical imaging accuracy should include details on the computational processes, input data types, and validation methodologies used to ensure reliability.

Additionally, the EPO places high importance on avoiding overly broad claims that do not directly relate to a technical field. Patent examiners scrutinize applications to ensure that AI inventions contribute to solving technical problems rather than merely automating human tasks or processing abstract data without a clear technical benefit. By requiring explicit disclosure of technical aspects, the EPO aims to prevent patents from being granted for inventions that do not advance technology in a meaningful way.

Comparative Analysis of Patentability for AI Inventions Patentability Requirements

Each jurisdiction applies distinct criteria:

• Brazil: AI-based inventions must demonstrate a “technical effect,” solving a problem beyond standard calculations or data manipulation. Software “per se” is unpatentable, requiring AI implementations to present industrial applicability and innovation.
• United States: The USPTO’s two-prong approach filters out abstract ideas, requiring AI inventions to integrate a practical application that provides a technical improvement.
• Europe: The EPO mandates a “specific technical purpose,” detailed technical disclosures, and reproducibility to ensure AI inventions advance technological progress.

Inventive Step and Reproducibility

• Brazil: Focuses on the technical effect and industrial applicability of AI.
• United States: Links inventive step to the two-prong eligibility test, ensuring AI contributes to specific technological advancements.
• Europe: Prioritizes reproducibility and a well-defined technical purpose, requiring detailed descriptions of AI algorithms and training data characteristics.

AI patent applicants should be aware of these distinctions to tailor their patent filings accordingly. Brazil and Europe emphasize technical contributions with an industrial or reproducible application, while the USPTO requires a demonstrable technical improvement in an applied field.

Conclusion

While Brazil, the US, and Europe have distinct approaches, their ultimate goal is to encourage AI innovation that provides tangible technical solutions. Brazil and Europe emphasize technical effect and industrial applicability, with the EPO placing additional focus on reproducibility. The USPTO’s two-prong test filters out abstract AI inventions, ensuring only those with functional applications receive patent protection.

For AI inventors seeking patent protection across multiple jurisdictions, applications should emphasize:

• A clear technical effect solving a specific problem.
• Integration into a defined technical field.
• Detailed algorithmic structures and data characteristics (particularly for EPO filings).
• A focus on measurable technical improvements that demonstrate advantages over prior technology.

By aligning applications with these criteria, innovators can enhance the likelihood of securing patent rights in Brazil, the US, and Europe. Strategic planning and jurisdiction-specific adaptations in patent applications will be critical for maximizing AI patent protection globally.

Disclaimer – The views expressed in this article are the personal views of the authors and are purely informative in nature.