Part 5: Freedom to Operate in Software, AI, Standards, and International Markets

The Freedom to Operate (FTO) Deep Dive: Questions Your Company Should Be Asking

This is Part 5 of a six-part series on freedom-to-operate analysis.  The series walks through what an FTO is, why it matters, how it works, what to do with the results, how it changes for software, AI, and international products, and the most common ways it goes wrong.  See the full table of contents here.

Part 5: Freedom to Operate in Software, AI, Standards, and International Markets.

An FTO is only as good as its scope, and certain product categories make scoping harder than usual.  Software and AI introduce moving-target terminology and aggressive filing activity.  Standards-essential patents create infringement that can’t be avoided.  International manufacturing and sales pull in multiple jurisdictions with different rules.  This post covers what changes for each of these and what to do about it.

How Does FTO Change for Software and AI?

The basic FTO process is the same for software and AI patents as for any other technology.  You define scope, search the relevant patent landscape, triage results, run claim-by-claim infringement analysis on the survivors, do claim construction informed by the specification and prosecution history, and produce a written opinion.  The substantive doctrines (literal infringement, DoE, PHE, and indirect infringement) all apply the same way they do everywhere else.  What changes is mostly at the edges of the process: how the search is structured, how the claim language tracks (or doesn’t track) the actual technology, and how often you need to refresh the analysis.  A few things in particular make software and AI FTOs more challenging than the same exercise for, say, a mechanical device.

For all software (including AI), the Supreme Court’s 2014 Alice decision made many software claims vulnerable to subject-matter eligibility challenges under 35 U.S.C. § 101.  That means a software FTO can sometimes lean more heavily on invalidity arguments than a hardware FTO would.  But the law on Section 101 is genuinely unsettled and is constantly changing, and relying on it alone is riskier than layering invalidity arguments alongside non-infringement positions.

For AI specifically, three additional things make FTO more challenging:

Terminology Drift.  What is called “a transformer” today might be called “an attention-based encoder-decoder architecture” in a 2017 patent.  Searches that rely on current terminology miss older patents using older language.

Pace of Change.  The technology is evolving faster than claims are drafted.  Patents filed in 2020 may use terminology that is already obsolete by the time they issue.

Active Patenting.  Major AI labs and platform companies are filing aggressively.  Static one-time searches age out very quickly.

For AI-heavy companies, a single pre-launch FTO probably isn’t enough.  The right approach is ongoing landscape awareness, with named-competitor watch lists refreshed regularly.  Continuation practice means a competitor’s patent portfolio can grow new claims at any time, and you want to see them as they emerge.

What About Standards-Essential Patents?

If you build something that implements a published standard (Wi-Fi, 5G, H.265 video, USB), there are patents declared “essential” to that standard.  By definition, if you implement the standard, you infringe those patents.

The good news is that the standards bodies require members to license those patents on FRAND terms (fair, reasonable, and non-discriminatory).  The question for standards-implementing products usually shifts from “Can we avoid these patents?” (you can’t) to “How do we get a FRAND license at a reasonable rate?”  This is a different exercise from a typical FTO.  The work focuses on contract negotiation and rate-setting.  Infringement avoidance plays a smaller role because, by definition, you can’t avoid the patents.  The legal framework is also still developing, with courts and standards bodies actively reshaping what FRAND means and what remedies patent owners can pursue when negotiations fail.

For products that implement a standard, the FTO question shifts from “what patents do we infringe?” to “what licenses do we need, and on what terms?”  That’s a different exercise, and worth scoping separately from the rest of the FTO work.

What Changes When We Manufacture or Sell Internationally?

Patents are territorial.  A U.S. patent only covers infringement in the United States.  So if you make a product in Taiwan, ship it to Germany, and sell it to a customer in France, you’re potentially dealing with patents in three different jurisdictions, each with its own rules.

Different Jurisdictions Construe Claims Differently.  The same claim language can mean different things in the U.S., Germany, and Japan.

Doctrine of Equivalents Varies Internationally.  Some jurisdictions are more aggressive than the U.S. about reading claims broadly; some are more restrictive.

271(g) Exposure. Importing products into the U.S. that were made abroad using a patented process can infringe U.S. patents. So even if all your manufacturing happens overseas, if the manufacturing process infringes a U.S. process patent, you can still get sued in the U.S. for selling the resulting product.

Prioritization Matters.  A truly global FTO covering every jurisdiction is rarely cost-justified.  Most FTOs focus on the major commercial markets where the product will be sold (typically the U.S., the EU, and one or two key Asian markets).  Manufacturing jurisdictions also matter if your process could itself be patented.  The right scope is a business decision based on where the actual revenue and exposure live.

For most companies, the right move is to define the scope of the FTO by where you actually do business.  A product sold only in the U.S. gets a U.S.-only FTO.  A product sold globally needs at least the major commercial markets covered.

The basic FTO framework holds across software, AI, standards-essential patents, and international markets, but the application varies.  Software and AI add complications around subject-matter eligibility, terminology drift, and the pace of new filings.  Standards-essential patents change the question from avoidance to FRAND negotiation.  International products require multi-jurisdictional analysis with different claim construction and DoE rules across borders.  The common thread is that scope definition gets harder, and getting it wrong creates blind spots that the opinion won’t catch.

Part 6 closes the series with the most common ways FTOs go wrong, who should do yours, and how to manage privilege around the opinion.