Ptab Ai Disclosure

The Patent Trial and Appeal Board processes over 1,500 inter partes review petitions annually, and AI is reshaping how both petitioners and patent owners prepare their cases. Prior art searches, claim construction arguments, and technical declarations are all being accelerated by AI tools — but the PTAB's unique procedural requirements create disclosure traps that general litigation AI workflows won't catch.

Finnegan — the largest IP-focused law firm in the U.S. — has been tracking AI developments across five patent tribunals, and their analysis reveals a fragmented landscape where disclosure obligations vary by proceeding type. The USPTO announced a voluntary Search Disclosure Declaration initiative in November 2025, signaling that AI-assisted prior art searches are on the agency's radar. For managing partners running patent prosecution and litigation practices, the question isn't whether to use AI at the PTAB — it's how to use it without triggering institution-stage problems.

PTAB Proceedings and AI: The Current Rules

The PTAB doesn't have a blanket AI disclosure requirement. But it does have strict duty-of-candor obligations under 37 C.F.R. § 42.11, requiring practitioners to provide information material to patentability in IPR and PGR proceedings. That duty intersects with AI in a specific way: if an AI tool identifies prior art references that are material to patentability, those references must be disclosed regardless of how they were found. The USPTO's November 2025 memorandum introducing voluntary Search Disclosure Declarations adds a new dimension. Petitioners in IPR and PGR proceedings can now voluntarily submit an SDD describing their search methodology — and the Board will consider it as a favorable discretionary factor supporting institution. That's a carrot, not a stick, but it signals where the USPTO is heading. Firms that document their AI-assisted search methodologies now will be better positioned when voluntary becomes mandatory.

AI in Prior Art Searches: Efficiency vs. Disclosure Risk

AI-powered prior art search tools — Ambercite, PatSnap, Google Patents with semantic search — can process millions of patent documents in hours instead of weeks. That's a genuine competitive advantage in IPR proceedings where the quality of prior art often determines whether the Board institutes review. But the efficiency comes with a disclosure obligation that most firms haven't fully grappled with. When an AI search tool returns 500 potentially relevant references and your associate narrows it to 10 for the petition, the question becomes: did you have an obligation to disclose the other 490? Under the current rules, you must disclose information you're aware of that's material to patentability. AI tools that surface references you wouldn't have found manually arguably expand your awareness — and therefore your disclosure obligations. The practical solution is to document your search methodology, including AI tool parameters, and apply consistent materiality standards to AI-generated results.

Director Squires and the Institutional Shift

In October 2025, Director John A. Squires announced he was assuming authority over all IPR and PGR institution decisions, in consultation with at least three PTAB judges. This centralization of institution authority changes the strategic calculus for AI-assisted petitions. Previously, institution decisions were made by three-judge panels with varying standards. Now, a single decision-maker applies consistent criteria across all petitions. For firms using AI to draft petitions, this means the quality bar for institution is more predictable but potentially higher. A petition drafted with AI assistance that contains superficially plausible but ultimately unsupported claim construction arguments will be evaluated by the same decision-maker who's seen that pattern before. The Director's centralized review also means that the voluntary Search Disclosure Declaration carries more weight — it's one person evaluating whether your search methodology supports institution.

Patent Prosecution and AI Inventorship Issues

Beyond PTAB litigation, the broader patent prosecution landscape creates additional AI considerations. The USPTO has maintained its position that only natural persons can be named as inventors — AI systems cannot be listed as inventors on patent applications. For PTAB proceedings, this means challenges to patents where AI played a role in the inventive process may raise novel invalidity arguments. If an AI system substantially contributed to the claimed invention, can the named human inventor satisfy the inventorship requirement? The Federal Circuit's Thaler v. Vidal decision said no for pure AI invention, but the boundary between AI-assisted and AI-generated invention remains legally undefined. Finnegan's analysis of patent disclosure requirements for AI inventions highlights the tension: patent specifications must provide enough algorithmic detail to demonstrate that the human inventor actually possessed the invention, not just the AI output.

Building a PTAB-Specific AI Compliance Protocol

Patent practitioners need a PTAB-specific AI workflow that addresses three unique risks. Prior art search documentation: Log every AI-assisted search, including tool name, search parameters, date, and results. Apply your materiality filter consistently and document your reasoning for excluding references. When the SDD becomes mandatory, you'll have the records ready. Petition and response drafting: AI-generated claim construction arguments must be verified against the patent specification, prosecution history, and relevant Federal Circuit precedent. PTAB judges read hundreds of petitions — they can identify formulaic, AI-generated analysis. Technical declarations: Expert declarations supporting or opposing institution should never be AI-drafted without substantive expert involvement. A declaration that reads like a language model's output rather than an expert's opinion will undermine credibility at institution and trial.

The Bottom Line: The PTAB's duty-of-candor obligations interact with AI in ways that general litigation rules don't address. The USPTO's voluntary Search Disclosure Declaration initiative is a preview of mandatory requirements to come. Patent practitioners should document AI-assisted search methodologies now, verify all AI-generated claim construction arguments against the prosecution history, and ensure expert declarations reflect genuine expert analysis rather than AI-generated boilerplate.