BYTES: Emotional Perception and AI Patents

Episode Summary

A landmark UK Supreme Court ruling could change how AI businesses protect their innovations. In this episode of “Bytes,” Skadden’s Maximilian Willis breaks down the Emotional Perception AI decision and its implications. The case centered on whether an artificial neural network, designed to recommend music based on emotional similarity to other music assessed through both semantic and musical properties of the music concerned, was excluded from patentability as a “computer program as such.” The Supreme Court moved away from the longstanding Aerotel test, adopting the European Patent Office’s “any hardware” approach. Because the ANN ran on a computer, it cleared the initial patentability hurdle, though the case returns to the UK IPO for further consideration. For AI businesses, the ruling raises an important question — could their systems now qualify for formal patent protection? 

Voiceover (00:01):
Welcome to “Bytes” from “SkadBytes,” jargon-free, bite-size insights from Skadden’s IP and tech team on the key issues shaping the tech landscape.

Maximilian Willis (00:12):
Hello, I’m Max from the IP and technology team here in Skadden, London. Here’s your Byte on the Emotional Perception AI UK Supreme Court decision, what it says and why it matters. The case concerned Emotional Perception’s artificial neural network, or ANN, a tool for recommending music based on emotional similarity to other music assessed through both semantic and musical properties of the music concerned. The question for the UK Supreme Court, which has been a key issue in patenting AI inventions in the UK to date, was whether the ANN tool was excluded from patentability as a “computer program as such.” Now, this is a specific exclusion from patentability under section 1(2)(c) of the Patent Act 1977. For nearly 20 years, the UK courts have applied a four-stage framework, known as the Aerotel test, to work out whether this exclusion for a “computer program as such” applies.

(01:15):
This test involved:

• First, properly construing the claim. In other words, working out what the invention actually is.
• Second, identifying the actual contribution that the invention makes to the state of the art.
• Third, asking whether that contribution falls within an excluded subject matter, in this case a “computer program as such.”
• And finally, checking whether the contribution is technical in nature.

Now, the UK Supreme Court decision here was very significant. First, it moved away from this longstanding Aerotel test, bringing UK practice broadly into line with the approach of the European Patent Office, or EPO. Now, this is important because it makes it easier, at least at the outset, for AI-related inventions to clear the exclusion hurdle.

(02:03):
So what’s the approach now? The Supreme Court adopted the EPO’s “any hardware” approach. This essentially means that at this initial stage, patent applications will not be rejected as a “computer program as such” where the claim involves the use of hardware. Now, the obvious example is a computer. So applying this to the Emotional Perception ANN. Second, the Supreme Court held that a “computer program” must be construed broadly and that an artificial neural network, in its trained state, falls within this definition. However, because the ANN ran on a computer, which is of course, as we mentioned, hardware, the application of the “any hardware” approach. While the ANN was a computer program, it was not a computer program “as such” and so was not excluded at this initial stage. The Supreme Court did not give an assessment of the ANN under the next two stages of the EPO’s approach, being first a determination of which features of the claim contribute to the invention’s technical character, and second, an assessment of whether those features are novel and involve an inventive step.

(03:10):
These questions were instead returned to the UK IPO, which has yet to make its determination. So this is not the end of the matter. Emotional Perception AI’s patent application has now returned to the UK IPO for further consideration of:

• Which features of the invention contribute to its technical character; and
• Whether the invention is novel and involves an inventive step.

The practical effect here could be significant. Companies that have leaned heavily on trade secrets or copyrights to protect AI or machine learning systems, both of which can be very awkward to enforce, may now find formal patent protection a more realistic option for AI tools characterized as computer programs. For the wider industry, the decision is a meaningful boost to legal certainty around AI innovation, particularly in the UK, and a signal that the patent system is adapting, however gradually, to the technologies it is increasingly being asked to accommodate. In short, if your business is developing AI systems, it is worth revisiting whether aspects of those systems could now be protected through patents.

Voiceover (04:21):
Thanks for listening to “Bytes.” Be sure to subscribe for more tech insights. Additional information about Skadden can be found at skadden.com.