On August 4, 2025, the USPTO’s Deputy Commissioner for Patents (Charles Kim) published a document entitled Reminders on evaluating subject matter eligibility of claims under 35 U.S.C. 101 which you may see here (archived here). I think the document may be of some help for applicants facing 101 rejections.
A first thing to note is that the “Reminders” document directs itself only to Tech Centers 2100, 2600 and 3600. One of my AI-related section-101 headache cases is in Art Unit 1685 (Nucleic acid assay, Combinatorial and Computational Chemistry, Combination Biology and Computer Science). I plan to cite this Reminder to my Examiner, but I anticipate that my Examiner will refuse to give it any mind since the Reminder does not say that it is directed to Tech Center 1600.
The Reminder has a chance, I think, of helping applicants a little bit.
We are all accustomed to the way it seems to work with Section 101.
Automatic rejection. It seems pretty much automatic nowadays that if “in a computer” appears in a claim, there will be a Section-101 rejection. It is the same thing for computer software. And absolutely always it happens if, in a text search, the words “neural network” or “AI” appear anywhere in the patent application.
The “reminder” says no, it should not be automatic to do this. The “reminder” says to do this only if a preponderance of the evidence says the rejection is appropriate. The section-101 rejection should not be imposed if the likelihood of success is 49% or less.
Mental process grouping. We have all seen the fairly standard approach these days for rejecting a method claim under Section 101. The Examiner avoids considering the claim “as a whole” despite the law requiring exactly that. The Examiner dissects the claim into a first part and one or more second parts. The first part is some grouping of steps that are called the “abstract idea” part. Normally what is also stated is that everything set forth in this first part can be performed in the human mind. The other parts are dismissed as “mere insignificant post-solution activity” or as insignificant parts that fail to bring the abstract idea back into the world of non-abstract things. The claim might say “taking the results from the previous step and chiseling them into stone and dropping them into the ground, generating a 3 on the Richter scale” and still this would be dismissed, without support, as supposedly being “mere insignificant post-solution activity”.
And later after you respond to that rejection, it is completely routine that absolutely every argument you have made is unpersuasive. No explanation why, but unpersuasive.
The “reminder” reminds the Examiner that the Examiner cannot just arbitrarily construct the “mental process grouping”. The Examiner is reminded, for example, that it is inappropriate to include in this grouping any step that cannot, in fact, be performed mentally.
Analysis of the claim as a whole. So now we turn to the parts of the claim that are not in the mental process grouping. It is routine that the Examiner takes one of those parts (the inevitably “insignificant” part that fails to bring the claim back into the real world of non-mental-step activity) by itself. But the “reminder” reminds the Examiner that the Examiner is not allowed to do this. The Examiner is required to analyze the claim “as a whole”.
Improvements. One of the ways to escape the “abstract idea” line of attack, and the “mere mental steps” line of attack, is supposedly to point to a “Step 2A prong two” consideration such as an improvement in the functioning of the computer, or maybe even an improvement in the functioning of another technology. When I am responding to a Section 101 rejection, I often point to such an improvement. (Fortunately, many of the AI and neural network patent applications that I am handling these days have the good luck to contain an “experiments” section that explicitly shows such an improvement in the experimental results. The neural network runs a bit faster, or gets the right answer more often, etc.) Invariably the Examiner dismisses this part of my response as “unpersuasive argument” with no explanation why it is unpersuasive.
The “Reminder” reminds the Examiner that if an improvement is shown, this ought to lead to the 101 rejection being withdrawn.
Interestingly, the “Reminder” actually says the Examiner is supposed to look through the specification to see if maybe the applicant actually addressed this in the specification. So maybe the “experiments” section that shows some improvement ought to prompt the Examiner to have the self-control to refrain from imposing the section-101 rejection in the first place. Said differently, the applicant should not need to point to this in the specification in an argument against a rejection. Instead, the Examiner should go and look for it and should find it, in which case the Examiner should have the self-control not to impose the Section-101 rejection in the first place.
Even more interestingly, the “reminder” tells the Examiner that the specification does not even need to explicitly set forth the improvement. Suppose that it is the case that the improvement, even if never set forth at all in the specification, would nonetheless “be apparent to one skilled in the art”. If so, the Examiner should recognize this and should have the self-control not to impose the section-101 rejection.
Still even more interestingly, the “reminder” tells the Examiner that this “improvement” need not be recited in the claim. It is enough that this “improvement” be set forth only in the specification. (Or, see previous paragraph, maybe the “improvement” need not be in the specification at all. Maybe all we need is the fact that one skilled in the art would find it to be “apparent”.)