The Patent Office Just Opened the Door for AI Founders — But It Won’t Stay Open Forever

Authored by Babak Akhlaghi on June 14, 2026.  In September 2025, something unusual happened at the USPTO.

The new Director — John Squires — walked into his office during his first week on the job and issued a ruling that sent a clear message to patent examiners across the country: stop categorically rejecting AI patents just because they involve machine learning.

The case was DeepMind Technologies (Appeal 2024-000567). The Board had rejected the claims as abstract. The Director reversed them.

And in doing so, he told examiners something founders have been saying for years: you’re getting this wrong.

The Cost Asymmetry That’s Been Crushing Founders

Here’s the reality most AI founders face when filing a patent application.

You file. You wait. Months later, the examiner comes back with a § 101 rejection, asserting your invention is “abstract.” Not that isn’t novel. Not that it is obvious. Just… abstract.

If you’re a big company with deep pockets, you fight. You appeal. You wait years if you have to.

But if you’re a founder on a limited runway, you can’t afford that fight.

Fighting with the examiner on 101 is just brutal. Founders don’t have the funds to keep fighting on an abstract, wishy-washy rejection.

The patent approval seal matters. It tells investors: Don’t take my word for it that my idea is unique. The patent office put the Approved seal on it.

Without it, you’re asking investors to trust your judgment alone.

The Examiner Lottery: Why the Message Hasn’t Reached the Ground Floor

Here’s what should worry you about the current system.

Even after this ruling, some examiners haven’t gotten the message.

Practitioners are pointing examiners directly to the DeepMind case. They’re citing the USPTO memorandums. They’re doing everything right.

And the examiners still aren’t withdrawing the § 101 rejections.

There is still lack of training. There is still lack of application. Some of these examiners have an 80% rejection rate. Their allowance rate is significantly low.

Translation: you can do everything right and still draw an examiner who hasn’t adapted to the new guidance.

And if you appeal? The reversal rate on § 101 rejections sits around 20%.

That’s not a coin flip. That’s an uphill battle.

Why NOW Is the Right Time to File

But here’s the thing.

The pendulum is swinging toward founders right now.

The Director’s message in the DeepMind case wasn’t subtle. He criticized the Board for doing a terrible job in setting forth the rejection. He pointed out that they merely said the claim recites an abstract idea and nothing else in the claim saves it.

That’s not analysis. That’s a checkbox.

And the Director made it clear: that’s not acceptable anymore.

The USPTO has issued memorandums. Examiners are being reminded that § 101 rejections need to be supported by a preponderance of evidence. If the patent owner is 51% right, the rejection should be withdrawn.

The environment is more favorable than it has been in years.

The door is open. Walk through it now before it’s too late.

What You Need to Do Differently When Filing

If you’re sitting on an AI invention and wondering whether it’s worth filing, here’s what you need to know.

Tell the technical story in your application.

What was the technical problem? What’s the technical solution? What are the advantages of your solution? How does it improve the functionality of the computer or another technology?

Don’t just put this in the specification. Make sure it lives in the claims.

In the DeepMind case, the claims included a limitation about training a machine learning model on a second task “while protecting performance of the machine learning model on the first machine learning task.”

That one line, that technical improvement, was the difference between eligible and ineligible. This improvement was also recited in the specification.

Make sure the claim captures the ‘how’ from a technical computer implementation standpoint. Broad functional statements are not enough.

Here’s your checklist:

• Identify the technical problem clearly
• Describe the technical solution in sufficient detail
• Outline the advantages of your solution
• Capture the “how” in the claims, not just the specification
• Cite Federal Circuit precedent, not just USPTO guidance if you get a 101 rejection

That last point matters. USPTO guidance gets you through examination. Federal Circuit precedent protects you in litigation.

Consider Expediting Your Application

The pendulum is swinging in your favor right now.

But pendulums swing back.

Directors change. Policies shift. The window won’t stay open forever.

If you’re sitting on the sideline, this is probably a good time to file. And consider expediting it to get it through the patent office more quickly.

Expedited examination costs more upfront. But it gets you to a decision faster — while the current favorable environment is still in place.

That matters when the alternative is waiting two years and hoping the policy landscape hasn’t shifted against you.

What to Do If You Get a § 101 Rejection

Even in this favorable environment, you might still get a § 101 rejection.

Here’s what to do.

First, check the examiner’s statistics. Look at their difficulty level, their allowance rate, what art unit they belong to. This tells you how hard the fight will be.

Second, interview. Don’t just respond in writing. Pick up the phone. Have a strategy before you call. Prepare a primary position and a backup position. Try to reach agreement with the examiner.

Don’t be adversarial with the examiner. They are doing their job. Try to understand their point. Put yourself in their shoes.

Third, if you hit an impasse, have an honest conversation about appeal. The chances of success on appeal are low, around 20%. The cost is real. But if you want to win, you need to know what you’re walking into.

And if you do appeal, cite both USPTO guidance and Federal Circuit precedent. That gives you leverage later if the patent gets litigated.

The Warning: This Window Won’t Last Forever

The DeepMind ruling is real.

The USPTO memorandums are real.

The shift in examiner guidance is real.

But none of it is permanent.

The pendulum is swinging toward the founders. But it won’t stay there forever.

A new director could take a different stance. Supreme Court could issue a new ruling that muddies the waters further. The favorable environment you see today could disappear in six months.

That’s why the message is urgent: file now.

If you’ve been sitting on an AI invention, waiting for the right time, this is it.

The door is open.

Walk through it before it closes.