‘Just File It’: The Risks of Filing Provisional Applications as a ‘Conduit’
Issues with 'cover letter' and AI-generated provisional applications
Patent practitioners are often approached by prospective clients, particularly solo inventors or smaller startups, with what seems like a simple request: to act as a “filing conduit.” The client has a technical write-up, an unpublished paper, or a detailed description of an invention. They are not ready for a full non-provisional patent application (NPA), but they want to secure a filing date and “patent pending” status immediately by filing a provisional patent application (PPA).
The client may even acknowledge that the practitioner lacks the specific technical expertise to draft claims for that technology. They offer to sign any waiver, stating they only want the practitioner to format the pages and submit the application. The client agrees to assume all responsibility for the disclosure’s content and legal sufficiency. Is there really a green light?
This scenario—now amplified by the rise of generative AI tools used by inventors—presents significant perils for both the client and the practitioner. It tests the boundaries of professional obligations, specifically the duty of competence and the requirements for enabling disclosure.
Practitioner Obligations Under USPTO Rules
While a client’s desire for speed and low cost is understandable, it does not supersede a practitioner’s duties under the USPTO Rules of Professional Conduct.
Duty of Competence: The foundational rule is 37 C.F.R. § 11.101, which states: “A practitioner shall provide competent representation to a client. Competent representation requires the legal, scientific, and technical knowledge, skill, thoroughness, and preparation reasonably necessary for the representation.” Filing a patent application, even a PPA, is a legal act. A practitioner who is admittedly not qualified to assess the technology arguably cannot provide competent representation, as they cannot judge if the disclosure meets the statutory requirements of 35 U.S.C. § 112.
Limiting Scope of Representation: Practitioners can, in some cases, limit the scope of their work. 37 C.F.R. § 11.102(c) permits this “if the limitation is reasonable under the circumstances and the client gives informed consent.” The central question becomes whether limiting the scope to “mere filing” is reasonable. Is it reasonable to abdicate the core professional function of evaluating a disclosure for legal sufficiency? A client’s “informed consent” is a high standard; they must truly understand the catastrophic risk of a defective filing.
Duty of Diligence: 37 C.F.R. § 11.103 requires a practitioner to “act with reasonable diligence and promptness.” Filing a document without any substantive review, knowing it may be legally void, could be construed as a failure of diligence. The practitioner’s signature on the filing documents signifies their involvement as counsel, an act that carries professional weight.
Where the § 112 Failure Will Emerge
A deficient “conduit” filing does not typically cause an immediate, visible problem. The USPTO will accept the PPA and issue a filing receipt. The problem is a latent one that will prove fatal in two key scenarios:
Situation A: The “As-Is” NPA Filing and Direct § 112 Failure
In this situation, the practitioner files the deficient PPA. A year later, another attorney (or the original practitioner) files an NPA that is essentially a copy of the PPA, perhaps with claims added that rely only on the original, deficient disclosure.
The Consequence: The NPA will face a direct § 112 rejection for lack of written description and/or enablement. The Examiner will state that the specification (the original PPA content) does not adequately support the full scope of the claims.
The Outcome: If the application somehow issues, it is exceptionally vulnerable to invalidation in federal court or at the PTAB on the same § 112 grounds. The patent will be found invalid for failing to teach a person skilled in the art (PHOSITA) how to make and use the invention.
Situation B: The “Beefed-Up” NPA Filing and Lost Priority
This is a more common and insidious failure. After the deficient PPA is filed, a competent attorney (the original practitioner after intense study, or new counsel) drafts a “beefed-up” NPA. This new application adds the necessary technical detail, drawings, and examples to properly support the claims. The practitioner files this NPA and claims priority back to the original “conduit” PPA.
The Consequence: An Examiner finds a piece of prior art that was published after the PPA filing date but before the NPA filing date. This new art anticipates or renders the claims obvious.
The Fatal Flaw: The practitioner’s attempt to claim priority to the PPA will fail. The Examiner will rule that the PPA does not provide § 112 support for the claims. Therefore, the claims are only entitled to the NPA’s filing date. The intervening prior art (published in the 12-month gap) is now valid against the application.
The Outcome: The claims are rejected under § 102 (Anticipation) or § 103 (Obviousness). The entire value of the PPA filing is lost, and the invention’s patentability is destroyed by the intervening art. This same “priority chain” attack can be used by a competitor in an Inter Partes Review (IPR) to invalidate an issued patent.
The Reckoning: Who Triggers the Consequence?
For years, the practitioner’s “conduit” filing may lie dormant. The danger becomes real when the patent is asserted or challenged, and the defective priority claim leads to its invalidation. At this point, the client has suffered a tangible, catastrophic loss, and they will seek recourse.
Two primary avenues of complaint could open, both aimed squarely at the practitioner:
The Legal Malpractice Lawsuit: The most direct threat comes from the client. The client, who has just lost their patent and potentially millions of dollars in a litigation or licensing deal, will file a malpractice suit in state court.
Who: The (former) client.
The Argument: The client will allege that the practitioner’s failure to provide competent counsel was the direct cause of their financial harm. The client’s new counsel will argue that the practitioner’s “waiver” is void. They will claim that the practitioner, as the licensed expert, had a non-delegable duty to review the disclosure for legal sufficiency and that allowing a client to “waive” this core competency is, in itself, a form of malpractice. The client will argue they could not have given informed consent because they are not a patent expert and relied on the practitioner to prevent such a fundamental error.
The OED Grievance: Separate from any civil lawsuit, the client can seek disciplinary action against the practitioner.
Who: The client, the client’s new attorney, or even the opposing counsel from the litigation that invalidated the patent can file a grievance. The USPTO’s Office of Enrollment and Discipline (OED) can initiate an investigation based on “a written submission from any source” alleging misconduct.
The Argument: The grievance will not focus on the financial loss, but on the practitioner’s violation of the USPTO Rules of Professional Conduct. The complaint will cite § 11.101 (Competence), arguing the practitioner knowingly undertook a task for which they lacked the technical skill. It will also cite § 11.103 (Diligence), arguing that “mere filing” without substantive review is a failure of diligence. The OED’s investigation will focus on whether the practitioner upheld their professional duty to the patent system, a duty that cannot be waived by a client.
In both scenarios, the “waiver” signed by the client is unlikely to save the practitioner. Instead, it may be presented as Exhibit A: evidence that the practitioner knew they were incompetent to handle the matter but proceeded anyway, placing their client’s patent rights in jeopardy.
Analysis: Weighing the “Conduit” Role
While accommodating a client’s request is a goal, the risks associated with this specific request are substantial.
Benefits:
For the client, the perceived benefit is speed, low cost, and the ability to mark a product as “patent pending.”
For the practitioner, it may seem like an easy way to satisfy a client’s immediate demand and maintain the business relationship.
Challenges and Risks:
The Illusion of Protection: This is the most significant danger. The client receives a filing receipt and believes they have secured a valid priority date. If the PPA disclosure is later found to be non-enabling or lacking written description, it cannot support a subsequent NPA. Any intervening prior art published after the PPA filing date could destroy the patentability of the invention. The “patent pending” status would have been worse than useless; it would have provided a false sense of security.
The Ineffectiveness of Waivers: A waiver is not a magical shield. When the client’s patent rights are lost, they will almost certainly challenge its validity, arguing they could not have given truly informed consent because they could not appreciate the technical legal standard of enablement they were supposedly “waiving.”
Abdication of Enablement Review: Leaving the § 112 analysis to the client is a severe gamble. A client, even a technically brilliant one, rarely understands the legal distinction between a technical description and an enabling disclosure that teaches a person skilled in the art (PHOSITA) how to “make and use” the invention without undue experimentation. The practitioner is paid for this exact legal judgment.
The Emerging AI-Drafting Risk: A related, and perhaps more immediate, issue is the rise of AI-assisted drafting. Practitioners without the requisite technical background may be asked to file specifications drafted by inventors using generative AI tools. These AI-generated disclosures can appear thorough, yet may contain subtle but critical “hallucinations” or lack the specific technical detail required to meet the § 112 enablement standard. This places the practitioner in the same compromised position: filing a document they cannot competently vet.
A Devil’s Advocate Perspective on Risk
On the other hand, for the purpose of academic discussion, a “devil’s advocate” position might hold that the risks, while real, can be managed.
First, it could be argued that “competence” under § 11.101 is not a static, binary attribute. For a practitioner with access to a deep bench of technical colleagues or consultants, or with sufficient time to study, becoming “competent enough” to review a disclosure is an achievable goal, provided the technology is not in a completely alien field (e.g., software versus pharmaceuticals). From this viewpoint, the practitioner’s primary failure is not filing as a “conduit,” but filing without a deep review—a step many practitioners would never skip.
Even in a pure “conduit” scenario, a practitioner might rely on a combination of defenses, such as:
(i) achieving a state of “minimal competence” to spot glaring omissions;
(ii) using the engagement letter to explicitly forbid the conversion of the PPA “as-is” into an NPA (addressing Situation A); and
(iii) if a “beefed-up” NPA later loses its priority claim (Situation B), arguing that the added material was “new invention” anyway, thus justifying the new filing date for that subject matter.
These arguments would pivot on the practitioner’s “professional judgment” exercised at the time, given the client’s explicit waiver.
However, even proponents of this view would concede that these strategies are far from bullet-proof. They would, at best, form a potential line of defense and would not avoid the significant time and expense of a malpractice suit or OED inquiry.
AI-Generated Invention Disclosures vs. “Cover Sheet” PPAs
The “conduit” problem is not new, but generative AI adds a deceptive layer to it. Historically, a “conduit” PPA was often just a technical whitepaper or a presentation slide deck—a document that was clearly not a patent application.
Today, an inventor can use an AI tool to generate a 50-page document that looks like a patent application, complete with a “Detailed Description” and “Summary.”
This creates the same, if not a greater, risk for the practitioner. The core issues are practically identical:
Lacking Enablement: A “cover sheet” PPA often fails enablement due to the omission of technical details. An AI-generated disclosure may fail due to the fabrication of technical details. AI models can “hallucinate” plausible-sounding but technically non-functional steps, components, or data. A practitioner without the subject matter expertise cannot distinguish this fiction from fact, leading to a non-enabling disclosure under § 112 (and/or an “abstract idea” challenge under § 101).
Loss of Priority Date and Foreign Rights: Just as in Situation B, this enablement failure is a fatal flaw. When the flawed, AI-generated PPA is used as a priority document for a later NPA, it will be found insufficient to support the claims. The priority date is lost. Any intervening prior art becomes valid, potentially destroying domestic patentability. This loss is even more catastrophic for foreign rights. Most foreign patent offices require “absolute novelty” and lack the one-year grace period found in the U.S. Any public disclosure by the client after the PPA filing—made under a false sense of security—will immediately forfeit all international patent rights once the priority claim is invalidated.
A False Sense of Security: This danger is magnified. The client, having seen a professional-looking, AI-generated document, is even more confident that their filing is secure. This false security encourages them to disclose their invention publicly or engage in commercial activities. When the priority claim is later invalidated, the client’s reliance on the “patent pending” status is exposed as having been worthless.
Ultimately, whether the document is a “back of the napkin” sketch, a technical whitepaper, or a polished AI-generated draft, the practitioner’s ethical obligation remains the same.
Filing a document without the technical competence to vet its contents for legal sufficiency under § 112 is an abdication of professional duty.
Potential Approaches for Practitioners
Instead of agreeing to act as a simple conduit, practitioners have more responsible options that protect both the client and their own license.
No “Rubber Stamps”: As a rule, do not act as a “filing conduit.” Practitioners must competently review anything they sign and file. Beware of “AI slop” coming from any source, including invention disclosures sent by long-familiar inventors who may be experimenting with new AI-based tools.
Decline the Representation: The most risk-averse and often most professional response is to decline the engagement. If the technology is genuinely outside the practitioner’s competence, 37 C.F.R. § 11.101 dictates this course. The practitioner can then refer the client to another firm with the requisite technical expertise.
The High-Standard Engagement Letter: If a practitioner still chooses to proceed, the engagement letter must go far beyond a simple waiver. For instance, it would explicitly state:
That the practitioner has not reviewed the disclosure for sufficiency under 35 U.S.C. § 112.
That the practitioner offers no opinion on whether the disclosure provides enablement or adequate written description for any future claims.
A clear warning that a deficient disclosure will result in the loss of the priority date, potentially causing an irretrievable loss of all patent rights.
That the representation is strictly limited to the administrative act of uploading the client’s documents to the USPTO.
Educate the Client: Often, the client simply does not understand the danger. It is the practitioner’s job to explain it. This includes counseling the client that a PPA is not a mere placeholder; it is the foundation of their patent right. A poorly written PPA is a foundation of sand, and the entire structure will collapse. It is far better to spend a small amount of extra time and money to have a competent practitioner review and augment the disclosure than to risk the entire invention on a flawed filing.
Step Back: Why Is This Happening?
These “conduit” requests do not occur in a vacuum. Understanding the inventor’s motivation is key to providing proper counsel. The drivers often include:
Financial and Time Pressure: This is the most common cause. The inventor may be trying to minimize costs, or they may be facing an imminent public disclosure, such as a trade show, a product launch, or an academic paper submission. The inventor may be making a quasi-public pitch in a few days. The PPA is seen as a fast, cheap way to get a “filing date” before this bar date.
A Breakdown in Education or Trust: The client may never have been properly educated on the critical function of a PPA. If they do not understand that the PPA is the disclosure that must support the final claims, they cannot appreciate the risk. This is sometimes compounded by a bad prior experience, such as paying high fees for a “weak” PPA that felt like a “cover sheet” filing anyway, leading them to view the service as a low-value commodity.
The “AI-Optimizer” Mindset: The rise of generative AI has created a new type of confidence. An inventor may use an AI tool to produce a document that looks polished and comprehensive. Convinced the AI has “optimized” the application, they believe the document is perfect and merely needs the “rubber stamp” of a licensed practitioner to be filed. This enthusiasm to have a “patent pending,” combined with a misunderstanding of AI’s potential for hallucination and its lack of true technical understanding, creates a strong conviction that a professional review is an unnecessary expense. This trend of inventors seeking a practitioner’s “rubber stamp” for what they believe is a complete, AI-generated disclosure is unlikely to disappear.
Understanding these underlying factors helps the practitioner shift the conversation from a “yes/no” on a filing request to a crucial counseling session about the real value of a patent application and the existential risks of a flawed foundation.
Conclusion: A Risk Not Worth Taking
The “filing conduit” request places practitioners in a difficult position, balancing client demands against fundamental professional duties. While it may seem like a low-cost, helpful service, it is a high-risk gamble with a client’s intellectual property and the practitioner’s professional standing.
The core of the problem is that a patent application is not a simple administrative form; it is a complex legal and technical document that serves as the basis for all future rights. A practitioner’s signature, even on a “conduit” filing, implies a level of professional review and competence. Maybe AI tools can provide that guarantee one day, but that doesn’t appear to be soon.
Ultimately, the potential for a catastrophic loss of patent rights, followed by a malpractice suit and an OED grievance, far outweighs the minor convenience of acting as a mere filing service.
The most prudent course is to adhere to the duty of competence: either perform a proper substantive review, associate with competent co-counsel, or decline the representation.
Selling a client a false sense of security is a service to no one. Allowing inventors to pass along or file “AI slop” is morally questionable, at best.
Disclaimer: This is provided for informational purposes only and does not constitute legal or financial advice. To the extent there are any opinions in this article, they are the author’s alone and do not represent the beliefs of his firm or clients. The strategies expressed are purely speculation based on publicly available information. The information expressed is subject to change at any time and should be checked for completeness, accuracy and current applicability. For advice, consult a suitably licensed attorney and/or patent professional.