Meta title: UK Supreme Court resets AI patents, aligns with G 1/19 Meta description: The UK Supreme Court refreshes software and AI patent rules, signaling closer alignment with EPO G 1/19. What it means for inventors, startups, and R&D. H1: UK Supreme Court reboots AI patentability and syncs with EPO G 1/19 The UK’s top court has delivered a significant course correction for the patentability of artificial intelligence and other computer-implemented inventions. In a judgment that will be closely read by software companies, research labs, and patent professionals, the Supreme Court has effectively “rebooted” how UK law treats AI-driven innovations by steering more firmly toward the European Patent Office’s (EPO) approach in G 1/19, the landmark 2021 decision on the patentability of computer-implemented simulations. While the UK has long had its own doctrinal tests for software and algorithmic claims, the ruling signals a more harmonized path that prioritizes demonstrable technical effect over labels like “software as such.” The result is a clearer roadmap for innovators: AI and simulation tools are not excluded merely because they run on a computer. What matters is whether their contribution produces a real, verifiable technical improvement—particularly one tied to the design or operation of a technical system or process. That framing will feel familiar to anyone prosecuting applications at the EPO, and it promises greater predictability for portfolios spanning the UK and continental Europe. H2: Why this matters now - AI spending and filings are surging: UK and European patent offices have seen a marked increase in AI-related applications, often touching multiple exclusions (mathematical methods, business methods, computer programs) at once. - Uncertainty was holding back protection: Applicants have struggled to predict outcomes in the UK when claims straddled software, simulations, or data processing. A more EPO-aligned lens offers greater legal certainty. - R&D investment rewards clarity: Early-stage companies and established R&D groups alike benefit when patentability criteria are transparent and consistent with European partners and investors. H2: The legal backdrop: from Aerotel to G 1/19 H3: The UK’s Aerotel/Macrossan framework For nearly two decades, the UK has relied on a four-step test, often called Aerotel/Macrossan, to assess excluded subject matter such as computer programs, business methods, and mathematical methods. In shorthand, the court asks: 1) Properly construe the claim. 2) Identify the actual contribution. 3) Ask whether the contribution falls solely within excluded subject matter. 4) Check whether the contribution is actually technical in nature. UK case law also developed helpful guideposts: - Symbian (Court of Appeal, 2008): A claimed improvement to the functioning of a computer itself can be technical. - Halliburton (2011): Computer-aided design and simulations aimed at real-world engineering can contribute technically when tied to a technical task or effect. H3: The EPO’s G 1/19 on simulations and technical effect In G 1/19, the EPO’s Enlarged Board of Appeal confirmed that computer-implemented simulations can be patentable if, assessed as a whole, they make a technical contribution. Importantly: - A simulation that assists the design or verification of a technical system or process can yield a technical effect. - The focus is not the abstract model or math alone, but whether the claimed method, when implemented, credibly solves a technical problem or informs tangible engineering decisions. - The “as such” exclusions remain, but they do not bar claims merely because mathematical methods or software are involved; the decisive question is the presence of a technical purpose and effect. H2: What the Supreme Court has reset While the UK has historically aimed for coherence with EPO practice, the Supreme Court has now given a firmer steer: - Emphasis on technical contribution: The decisive question is not whether an invention is “software” or “AI,” but whether the claimed contribution produces a real technical improvement—faster, more reliable, or more effective operation of a technical system; improved resource management; or the design/verification of a physical or industrial process. - Alignment with G 1/19 for simulations: Claims that model or simulate technical systems or processes, especially where they inform engineering choices or control parameters, are more clearly within the realm of technical subject matter than before. - Substance over form: Drafting that merely dresses up a business or informational outcome will not pass. Conversely, claims that pinpoint credible technical advantages, corroborated by implementation detail, stand a stronger chance. - No carte blanche for AI: The judgment does not open the floodgates. Pure data analysis, presentation of information, or abstract algorithms without a specific technical purpose remain excluded. The key is tying AI or ML to a concrete technical problem and solution. H2: Practical impact for AI and software innovators H3: Stronger prospects for simulation-driven R&D - Engineering simulations: Models used to design or verify behavior of mechanical parts, fluid dynamics, thermal management, or wireless signal propagation are now more squarely in the technical camp if claims articulate how the simulation achieves a verifiable design or control benefit. - Control systems and optimization: AI tools that tune parameters of a technical process (e.g., energy optimization in HVAC, vibration damping in machinery, or packet scheduling in networks) can demonstrate technical effect. H3: What still won’t fly - Abstract math or statistics: A clever model architecture or training heuristic that improves classification accuracy in the abstract, without a defined technical use or effect, remains vulnerable to exclusion. - Business and administrative logic: Recommender systems, pricing engines, or risk scoring pipelines that deliver non-technical outputs (e.g., marketing, finance, HR) are unlikely to qualify unless they also tangibly change the behavior of a technical system. - Mere data presentation: Better dashboards, visualizations, or alert formats—without changes in how a technical system operates—are typically excluded. H3: Machine learning claims that can clear the bar - Improving the computer itself: If an ML method measurably reduces memory footprint, latency, cache misses, or energy consumption at the hardware or OS level, those gains can count as technical improvements. - Technical sensing and signal chains: Claims tied to sensor fusion, noise suppression, or interference mitigation—where the ML method demonstrably improves physical signal processing—are more likely to be accepted. - Edge computing and deployment: Methods enabling on-device inference with quantized models, novel pipeline schedulers, or bandwidth-aware compression that maintains acceptable performance for a defined device class can indicate technical effect. H2: Drafting and prosecution: what to change now H3: Make the technical problem explicit - Start with a well-defined technical problem: e.g., thermal hotspot detection in a battery management system, or latency reduction in a real-time control loop. - Quantify the improvement: include metrics like reduced CPU cycles, lower packet loss, faster convergence, improved stability margins, or better mean time between failures. H3: Tie claims to a technical system or process - Anchor your claims to a tangible system: a network stack, a robotic actuator, a fabrication line, a power converter, or a medical imaging pipeline. - Include implementation details: data structures, memory layouts, timing diagrams, or control flow that drive the technical effect. H3: Use evidence and credibility - Provide experimental data, benchmarks, or simulations that correlate to real-world behavior. - Avoid hand-waving: generic “improves performance” assertions should be replaced with test setups, boundary conditions, and tolerances. H2: Portfolio strategy: UK, EPO, and beyond - Harmonize with EPO practice: Draft your UK and EP applications with the COMVIK/G 1/19 mindset—separate technical and non-technical features, and pinpoint the technical contribution in the problem-solution framework. - Stage claims: Consider including an independent claim that pairs the AI/simulation with its deployment in a technical system, and auxiliary claims directed to the computer-implemented method with clear technical constraints. - Coordinate timelines: If prosecuting in both the UKIPO and EPO, keep arguments and evidence aligned. Consistency on the articulated technical problem and effect can reduce friction and expedite allowance. H2: Interaction with inventorship: people, not machines The Supreme Court’s direction on patentability does not change the UK’s stance on inventorship. Following prior Supreme Court authority, only natural persons can be named as inventors on UK patents. AI systems cannot hold inventorship or ownership. Companies should: - Record human contributions to conception and reduction to practice. - Maintain lab notebooks and code repositories that evidence inventive input by named personnel. - Align employee IP agreements and assignment chains to avoid post-grant disputes. H2: Litigation outlook and UKIPO practice - Expect refined guidance: The UKIPO typically updates practice notes following major judgments. Applicants should watch for examiner guidance that mirrors the renewed focus on technical effect and G 1/19-consistent reasoning. - More predictable outcomes: Greater convergence with EPO principles should translate into more consistent examination and fewer surprises on appeal. - Claim construction remains critical: Courts will continue to scrutinize the “actual contribution.” Precision in how the invention is framed—what it does to a technical system and how—will often determine success. H2: Examples: when is an AI invention likely patentable? H3: Likely in-scope - A neural-network-based controller that stabilizes a drone in gusty wind by adapting PID coefficients in real time, reducing oscillation amplitudes by 30% in wind-tunnel tests. - A learned modem equalizer that reduces bit error rates on noisy copper lines without raising transmit power, validated across standardized channel models. - A simulation method that predicts turbine blade resonance under variable load and feeds back tuned damping parameters to the manufacturing process. H3: Likely out-of-scope - A deep learning classifier that labels emails as “urgent” versus “non-urgent” for task management, with no change to how any technical system operates. - A portfolio optimization algorithm that reallocates assets based on sentiment analysis, presented in a dashboard. - A generic recommendation engine that suggests products, unconnected to controlling or improving any technical process. H2: Key takeaways - AI is not excluded per se: What matters is a concrete technical purpose and credible technical effect. - Simulations can be patentable: When they inform or improve the design or operation of technical systems, consistent with G 1/19. - Draft for substance: Claims and specifications should highlight measurable, implementation-grounded improvements in a technical system. - Expect closer UK–EPO alignment: A welcome boost for cross-border portfolios and R&D planning. H2: Suggested featured image - The Supreme Court of the United Kingdom building in London, symbolizing the judicial reset on AI patentability. - URL (Commons): https://upload.wikimedia.org/wikipedia/commons/4/49/Supreme_Court_of_the_United_Kingdom_2014.jpg - Alternative (EPO Munich HQ for EPO alignment context): https://upload.wikimedia.org/wikipedia/commons/5/5e/European_Patent_Office_Munich.jpg FAQs Q1: Does the Supreme Court ruling make it easier to patent AI in the UK? A1: It makes outcomes more predictable when your AI invention delivers a genuine technical effect. If your claims show how the AI improves a technical system—such as lowering latency in a control loop, reducing power consumption, or enabling more accurate sensor processing—they stand a stronger chance. Purely abstract data analysis or business logic, however, remains excluded. Q2: Do I need to name the AI system as an inventor on my UK application? A2: No. UK law requires inventors to be natural persons. You should name the human contributors who conceived of the inventive concept and ensure proper assignment to the applicant company. This inventorship rule is separate from the question of whether the subject matter (such as AI or simulations) is patentable. Q3: How should I draft AI and simulation patent applications after this ruling? A3: Lead with the technical problem and the measurable technical improvement. Tie the invention to a specific technical system or process, explain implementation details that enable the effect, and provide credible evidence (benchmarks, experiments, or validated simulations). Align your UK filings with EPO practice (G 1/19 and COMVIK), distinguishing technical from non-technical features and focusing on the technical contribution.