Loading issue research
Patent law should recognize human-directed AI inventorship, fix AI-era eligibility rules narrowly, and keep patent review open so trolls gain no new leverage over startups.
Verification Status
AI-researched, unverifiedLast Reviewed
Jul 5, 2026
Cited Sources
15
Check how the claim was researched, how confident it is, and the evidence behind it.
Three separate mechanisms, addressing three separate live fights, in the order courts, the Patent Office, and Congress are moving on them: (1) codify in statute what courts and the Patent Office have already settled through case law and guidance, that only a natural person can be a patent inventor under ordinary conception doctrine, the operative standard the Patent Office's 2025 guidance applies by treating AI as a tool rather than a potential joint inventor; (2) resolve Section 101's uncertainty for AI and software claims narrowly, anchored to the technical-improvement standard the Federal Circuit established in Recentive Analytics v. Fox Corp., rather than the broader rewrite the pending Patent Eligibility Restoration Act currently proposes; (3) reject the PREVAIL Act's restriction on who may petition for inter partes review, and substitute a narrower fix, a bar on duplicate petitions re-litigating grounds already rejected against the same patent, that answers the same abuse concern PREVAIL cites without reducing standing for nonprofits, open-source defense groups, and future defendants. None of these three mechanisms revisits whether or how AI companies should pay to train on copyrighted material; that question, and this platform's answer to it, belongs entirely to AI-10.
This position is procedural, not normative: it does not adjudicate whether patents are good policy or whether an algorithm deserves legal standing. It sets specific institutional rules, who counts as an inventor, what counts as patent-eligible, who may challenge a patent, so that the rules match how invention and litigation work in 2026 rather than how they worked when Section 101's current test was written or when the PTAB's standing rules were set.
Patent rules should track who or what supplies the inventive judgment behind a claim, and any change to how patents are challenged or enforced should be judged against the litigation landscape that exists today. The narrow claim: a natural-person-only inventorship rule and a technical-improvement eligibility standard are the correct baseline for AI-assisted invention, and no enforcement reform should reduce challengers' access to patent review in a market where non-practicing entities already file most tech-sector suits.
Primary — Research, Innovation, and Collaboration. The mechanism is direct: fixing Section 101's uncertainty for AI and software claims, and putting inventorship on a statutory footing instead of rescindable guidance, determines whether the R&D this value wants to encourage produces protectable, investable intellectual property. Right now the slowest-moving technology centers at the Patent Office are the AI- and software-heavy ones, which is this value's own investment thesis running into an administrative bottleneck.
Secondary — Inclusive Growth and Economic Development. "Startup-friendly policy" is the operative phrase: preserving broad inter partes review standing (Proposal 3) is the concrete floor that keeps fast-moving, cash-constrained companies from being priced into patent-troll settlements they can't afford to litigate against.
Acknowledged tension — the same Inclusive Growth and Economic Development value pulls the other way for a different constituency. An individual inventor or small biotech firm with a legitimate patent is exactly the kind of underdog innovator this value should also protect, and that inventor's patent stays open to serial challenge from better-funded incumbents under this issue's own preferred approach. This isn't resolved by pretending both groups get everything. Proposal 4's bar on duplicate petitions is the load-bearing answer for that constituency, the same role Proposal 5's safe harbor plays for smaller developers in AI-10.
This is not a partisan fight. PERA, PREVAIL, and RESTORE all carry matched Republican-Democrat sponsor pairs: Tillis and Coons in the Senate on PERA and PREVAIL, Cotton and Coons in the Senate on RESTORE, Kiley and Peters and separately Moran and Dean in the House. PREVAIL's predecessor cleared the Senate Judiciary Committee 11-10, a margin that reflects an industry-coalition split, not a party-line one. Opposition runs the same way: the Electronic Frontier Foundation and the startup coalition Engine Advocacy oppose the package from a technology and public-interest angle, while the Heritage Foundation has separately argued that current PTAB rules already let well-resourced incumbents outlast small patent holders, an argument for strengthening patent-holder protections. Neither major party's leadership has staked out an institutional position on any of these bills; the coalitions are industry interest groups formed independently of party leadership.
The Innovation Party's delta is a refusal to treat "the patent reform package" as a single yes-or-no vote the way its sponsors and opponents both currently frame it. Inventorship, eligibility, and enforcement are three different questions with three different answers here: support the settled inventorship rule and a narrower eligibility fix, reject the standing restriction, and support a substitute enforcement fix that answers the abuse concern without the standing cost. That's not a synthesis of two existing sides, since the current sides are organized by industry rather than ideology; this issue's delta rejects the premise that these bills should be voted on as one bundle at all.
The strongest good-faith objection comes from the same direction as Heritage's institutional critique: broad third-party standing to petition for inter partes review has, in practice, let well-resourced incumbents file petition after petition against the same small inventor's or university's patent, a pattern critics, including a former Federal Circuit chief judge, have called PTAB acting as a "patent death squad." If that's the dominant practical effect of broad standing, then defending it primarily protects the same large incumbents this issue elsewhere treats as the target of patent-troll suits, not the small inventors this issue claims to serve, and PREVAIL's narrower standing rule is a more honest match for who the tool should serve.
The party's answer is to separate the abuse from the tool rather than defend unlimited petitioning. Proposal 4, barring duplicate petitions that re-litigate grounds already rejected against the same patent, answers the "death squad" pattern directly: no patent holder, large or small, has to survive the identical argument twice. What Proposal 4 does not do is remove standing from the parties Heritage's critique isn't about, the nonprofits, open-source defense groups, and future targets whose only realistic point to challenge a bad patent is before they've been sued, when the fight is still cheap enough to bring. The scale of the two problems settles which one the standing rule should be built around: non-practicing entities filed 88.3% of the high-tech patent lawsuits Unified Patents tracked in district courts in the second quarter of 2024, a market broad standing exists to check; there is no comparably documented pattern of biotech and university patents being exhausted by serial incumbent challenges at that scale. The position holds: cap the specific abuse with Proposal 4, leave the checking mechanism itself intact.
Startups and small tech companies targeted by non-practicing-entity suits bear an acute, concentrated cost today, and this issue's rejection of PREVAIL's standing restriction (Proposal 3) is built to protect them, at the cost of leaving PTAB open to the incumbent-abuse pattern named in the Steelman above. That cost is not left unaddressed: Proposal 4's bar on duplicate petitions is this issue's specific answer for the individual-inventor and university population absorbing it, not a residual gap.
Individual inventors, university tech-transfer offices, and small biotech firms bear a cost under continued broad third-party standing: their patents stay open to challenge from parties with no direct stake in the outcome, including incumbents with the resources to file repeatedly. This issue accepts that cost because the documented alternative, PREVAIL's fix, would remove the same challenge tool from the population currently absorbing the larger, better-measured harm, non-practicing-entity litigation against startups. Proposal 4 is the deliberate mitigation this position builds in for exactly that population.
Patent applicants in AI, software, and diagnostics who wanted PERA's broader Section 101 rewrite bear a cost from this issue's narrower alternative: some claims that would clear examination under PERA's text will still fail the technical-improvement standard this issue proposes instead. This issue accepts that cost to avoid PERA's unmodified-gene loophole and near-automatic software eligibility, the specific problem the Alice and Myriad decisions were written to stop.
Non-practicing entities and the litigation-finance arrangements that back them lose the leverage PREVAIL and RESTORE would hand them. That is not treated here as a regrettable side effect; it is Proposal 3's intended result.
Turn frustration into useful pressure.
If this position misses evidence or a lived consequence, challenge it. If it holds up, help test it locally and connect it to the issues around it.