USPTO Reverses Course on AI Inventorship, UK Supreme Court Rewrites Software Patent Law

The global patent landscape for artificial intelligence underwent a seismic shift this year. Two pivotal rulings have fundamentally altered how companies must approach AI-related intellectual property strategies.

USPTO Abandons Biden-Era AI Guidance

On November 28, 2025, the USPTO issued revised inventorship guidance that rescinded the February 2024 framework requiring "significant contribution" analysis for AI-assisted inventions. The change eliminates AI-specific heightened standards and clarifies that traditional conception principles apply universally.

The prior guidance demanded that "a natural person must have significantly contributed to each claim in a patent application." Patent prosecutors spent months documenting human prompts, analysing AI outputs, and building inventorship arguments around the three-part Pannu test from joint inventorship doctrine.

That framework is gone. The November guidance states that joint-inventor analysis applies "only when determining whether multiple natural persons potentially qualify as joint inventors and cannot be applied to hold AI systems as 'joint inventors.'"

Thaler v. Vidal remains controlling law: AI systems cannot be named inventors. But the practical burden has shifted. Instead of proving significant human contribution to each claim element, applicants need only establish that a natural person conceived the invention using traditional standards developed in decades of Federal Circuit precedent.

The reversal creates immediate tactical advantages for patent prosecutors. Standard conception analysis suffices, the same framework that governed patent law before generative AI existed.

UK Supreme Court Overturns Twenty Years of Patent Practice

The UK Supreme Court delivered an even more dramatic reset. In Emotional Perception AI Limited v. Comptroller General of Patents on February 11, 2026, the Court unanimously abandoned the Aerotel four-step test that had governed software patent eligibility since 2006.

The case involved an artificial neural network that recommends media files based on emotional similarity rather than genre classification. The UK Intellectual Property Office rejected the application as a "program for a computer…as such" under the Aerotel framework, which required identifying the invention's "actual contribution" and assessing its technical character.

The Supreme Court held that Aerotel "wrongly collapses the distinct EPC requirements" by importing novelty analysis into the threshold eligibility inquiry. The Court adopted the European Patent Office's "any hardware" approach from G1/19: if a claim involves any physical hardware, it clears the computer program exclusion.

Under Aerotel, patent disputes centred on whether an invention made a sufficient "technical contribution." Under the new framework, any claim requiring hardware passes the first hurdle. The real analysis moves to an "intermediate step" that filters technical contributions before assessing novelty and inventive step.

Practical Consequences for AI Patent Strategy

The USPTO guidance creates immediate procedural advantages but doesn't resolve deeper questions about AI patent eligibility under 35 U.S.C. § 101. The Supreme Court declined review in Recentive Analytics v. Fox in December 2025, leaving Federal Circuit doctrine controlling artificial intelligence subject matter eligibility.

The UK ruling provides clearer pathways for AI patent protection. Any neural network implementation necessarily requires computer hardware, automatically satisfying the eligibility threshold. The challenge becomes proving that non-technical features contribute to technical character.

But uncertainty remains. The Supreme Court remitted Emotional Perception for reconsideration under the new framework, meaning no definitive guidance yet exists on how the UKIPO will apply the "intermediate step" analysis in practice.

Strategic Implications

For multinational AI portfolios, these changes require immediate attention. US applications can now proceed under traditional inventorship standards without AI-specific documentation requirements. UK applications benefit from a lower eligibility threshold but face new analytical frameworks.

Cross-border coordination becomes critical. Companies should audit pending applications for inventorship consistency and prepare for heightened scrutiny in disputes involving AI-assisted inventions. Courts will likely demand solid evidence of human conception, even under relaxed standards.

The divergent approaches also highlight broader jurisdictional splits. While the UK aligns with EPO practice, the US maintains its own inventorship doctrine. South Africa remains the only jurisdiction recognising AI inventors, though this outlier position carries limited practical significance.

The USPTO's revised guidance takes immediate effect, but the full implications await testing in prosecution and litigation. The UK framework requires UKIPO implementation guidance, expected within months as the office processes the remanded Emotional Perception application.