The Intellectual Property Office of the Philippines (IPOPHL) has introduced updated rules to clarify how inventions developed with artificial intelligence qualify for patents. Released in July 2025, the guidelines define AI-assisted inventions and outline how patent examiners will evaluate them, addressing the increasing use of AI across sectors from pharmaceutical research to financial fraud detection.
AI already plays a central role in innovation. In banking, it authenticates users and detects suspicious transactions. In medicine, it processes genetic data to accelerate drug trials. In farming, AI-managed irrigation systems reduce waste while increasing crop yields. As AI becomes more integral to invention processes, patent authorities worldwide are revising their frameworks. The Philippines’ new standards aim to standardize evaluations of AI-generated work.
Defining AI-assisted inventions
The rules categorize AI-assisted inventions into two groups. First are core AI technologies—innovations that enhance AI systems themselves, such as new neural network designs or improved training methods. The focus here is on advancing the AI model or its underlying processes.
Second are applied AI technologies—solutions that deploy existing AI to address practical challenges in fields like medicine, logistics, or finance. In these cases, the innovation lies in the application rather than the AI itself. For example, an AI-controlled assembly line would qualify as applied AI, while a faster AI training algorithm would fall under core AI.
Technical contribution determines patent eligibility
Not every AI-assisted invention will receive patent protection. The guidelines require that such inventions meet standard patent criteria: originality, an inventive leap, and practical utility. Before these can be assessed, however, the invention must first qualify as patentable subject matter—a threshold where AI introduces complications.
In the Philippines, AI inventions are classified under computer-implemented inventions. To be patentable, they must demonstrate a technical contribution—a measurable improvement over existing solutions. Running an AI model or performing calculations alone does not suffice. The invention must resolve a specific technical challenge with tangible benefits, such as higher efficiency, greater precision, or enhanced safety in a physical system.
An AI system that optimizes urban traffic flow could meet these standards. A tool that simply organizes data without a clear technical advantage would not. The guidelines establish a five-step evaluation:
- Does the claim involve AI and computer implementation?
- Does it conflict with public order or ethics?
- Is it tied to a product or process under the IP Code?
- Does it involve non-technical elements?
- Does it retain technical substance after review?
If the answers confirm a technical effect, examiners proceed to evaluate novelty, inventiveness, and industrial applicability.
Read Also: Patent Attorneys Open AI Law Firm in Japan
Precision in claim drafting is also mandatory. Every term must be technically precise, unambiguous, and consistent with the invention’s description. Vague phrasing risks rejection, so applicants should structure claims as method or system descriptions with explicit features.
Human oversight remains essential
The guidelines explicitly state that AI cannot be named as an inventor or applicant on a patent. Philippine law mandates that inventors be human, as only individuals can enforce intellectual property rights. Even if AI generates an invention, the person who provided initial inputs—such as prompts or training data—must be listed as the legal inventor. Joint inventorship follows the same principle: all contributors must be human.
AI-generated industrial designs—whether created with human guidance or autonomously, can still be registered. However, the applicant must be a legally recognized entity. While AI tools are not prohibited in design creation, the final applicant must have the capacity to defend against infringement claims.
This requirement carries practical implications. Companies using AI to draft patents or designs must ensure human involvement is properly documented. Overdependence on AI without clear human input could weaken a patent’s validity if contested.
Legal and ethical concerns persist. Though the IP Code does not ban AI-assisted creation, other laws, such as those governing copyright or deceptive practices, may apply if AI-generated work infringes existing rights or misrepresents human authorship.
For inventors and businesses, the core principle is clear: AI is a valuable tool, but patents and designs still demand human oversight to satisfy legal requirements. The IPOPHL’s guidelines do not prohibit AI use, they establish boundaries for protecting its contributions.
The document requires applicants to disclose training datasets, algorithms, technical effects, and working examples in sufficient detail for replication.
These rules highlight a broader conflict in AI law. Governments must balance encouraging innovation with preventing legal disorder. AI speeds up discovery but challenges traditional notions of inventorship and creativity. The Philippines’ approach favors practicality: AI can assist, but human accountability is non-negotiable. Startups and researchers must invest in documentation and oversight from the outset, treating AI as a collaborative tool rather than a standalone creator. The guidelines do not hinder progress, they impose clarity where uncertainty once prevailed.
