AI Disclosure

1. AI Output Is an Unverified Draft

All AI output of the Service consists of unverified drafts. It may contain factual errors, cite non-existent literature or judgments, misinterpret the scope of claims, or omit key prior art.

The output does not constitute legal advice, a patentability opinion, an infringement assessment or any professional judgment; it shall not be used as the sole basis for any application, response, litigation or business decision.

2. Human in the Loop: The Signatory Is Always a Human

The final judgment on, revision of and signing of every document issued to external parties shall be carried out by a qualified patent attorney, patent agent or attorney-at-law, who shall be responsible for its content.

The user's duty of verification is neither reduced nor transferred by the use of AI tools. The USPTO's April 2024 guidance on the use of AI-based tools likewise makes clear that the person submitting a document must have personally reviewed and verified its content, and that "reliance on the accuracy of an AI tool" does not constitute a reasonable inquiry.

3. Limitations of Search and Citation

Some functions of the Service use web search, the Judicial Yuan judgment database or knowledge-base retrieval as aids. The coverage, currency and completeness of these sources are all subject to limitations, and they are not equivalent to a professional search database. In the functions that use web search (Prior-Art Search, Research & Competitive Analysis and Validity Assessment), the model itself generates search keywords from the content input by the user and sends them to Google Search; such keywords may reflect the technical features of the input content.

The knowledge-base source markers shown in AI answers (such as [KB-1]) can help trace the source passage, but the user must still check the original text; all cited statutory provisions, chapters of the Patent Examination Guidelines and judgment numbers must be verified by the user.

The results of Prior-Art Search shall not be regarded as a complete patentability search; the fact that no reference is found does not mean that no relevant prior art exists.

4. Confidentiality Classification

Input content will be transmitted to a third-party AI service for computation. The user shall determine, in accordance with the confidentiality rules of the organization to which the user belongs and the terms of engagement with clients, what content is appropriate to input.

Where unpublished invention content, priority documents or information that a client has designated as not to be disclosed externally is involved, please first confirm that the necessary authorization has been obtained.

Users are advised to de-identify and desensitize content before inputting it: replace identifying information such as applicant, inventor, client and docket numbers with codes, replace key formulations, parameters and business figures with placeholders or approximate ranges, and input only the passages relevant to the question. De-identification should be completed outside the Service and should not be performed using the AI functions of the Service.

5. Rights Risks Associated with Output

The Company does not warrant that AI output is original, nor that it does not conflict with the copyright, patent rights or other rights of third parties.

Under the practice of the various jurisdictions, artificial intelligence may not be named as an inventor; an inventor must be a natural person. The user shall ensure that the inventor designation in application documents complies with the requirements of each relevant jurisdiction.

6. Limitation of Liability

The Company shall not be liable for any damages arising from reliance on AI output; all other limitations of liability are governed by Section 6 of the "Terms of Service".