Stop press: UK Supreme Court rewrites the rules on patentability of Computer Implemented Inventions

The UK Supreme Court in today’s decision in Emotional Perception AI v Comptroller General [2026] UKSC 3 has brought about a seismic shift in the way that patentability of Computer Implemented Inventions is to be assessed in the UK. Computer programs “as such” are excluded from patentability under s. 1(2)(c) of the Patents Act 1977 (corresponding to Arts 52(2)(c) and 52(3) EPC) and for 20 years, the courts have followed the “Aerotel” test set out by the Court of Appeal in Aerotel Ltd v Telco Holdings Ltd [2006] EWCA Civ 1371; [2007] Bus LR 634; [2007] RPC 7. The “Aerotel approach” put the English courts at odds with EPO practice, which has followed a different approach to the issue, often referred to as the Comvik test, taken from the EPO's TBA decision T641/00.

The UK Supreme Court has now ruled that the Aerotel test should no longer be followed. The Court noted that this test was a legitimate product of its time; a time when it was not clear that the EPO’s Comvik test would endure. However, the Court noted that since the outcome of the Enlarged Board of Appeal case G1/19, which overtly endorsed the Comvik approach and implicitly endorsed criticism of Aerotel, this is no longer the case. Accordingly, the Court decided that the UK approach on this issue should be consistent with that of the EPO and G1/19, and remitted the case to the UKIPO Hearing Officer to apply the new standard.

A further interesting aspect of this case, as patent practitioners in the field of computer implemented inventions will appreciate, is that the Comvik approach necessitates considering whether individual features of an invention are technical (or non technical) as part of, or prelude to, inventive step assessment. It has been clearly demonstrated in the EPO how this consideration can be stitched into the EPO’s “problem-solution” approach to inventive step. It will now be incumbent on English courts to do the same. The Court dismissed arguments that by endorsing G1/19 (and thus, the Comvik approach), it would be necessary to replace the English courts’ established approach to inventive step under Pozzoli with a “problem-solution” style approach, and they overtly endorsed the Pozzoli test. They also discounted arguments that doing so would create a “Frankenstein monster” of a test. Nevertheless, patent practitioners, examiners and English courts will now have to forge new ground in figuring out how to adapt a “technical features only” filter for the Pozzoli test.

Courts have long struggled with the question of how to separate out the test for patent eligibility from tests for novelty and inventive step. With this decision, the UK Supreme Court has overtly sided with the EPO in concluding that it is more appropriate to import an assessment of the “technical” nature of claim features into an inventive step assessment rather than to import questions of inventiveness into an enquiry into patent eligibility. This aligns the UK and the EPO on the opposite side of the debate to the US which it could be said still includes aspects of obviousness/prior art analysis as part of how the 2-step Mayo/Alice patent eligibility test for Computer Implement Inventions established by the US Supreme Court is dealt with in practice.