On 2 July 2025, the Enlarged Board of Appeal of the European Patent Office (EPO) issued its landmark decision in case G 1/23 (Solar Cell Sealing Materials and Solar Cell Module), a ruling that has sparked a wide range of interpretations, from “nothing to see here”, to concerns that in Europe, we may now face an effective equivalence to the US’s on-sale bar. Although G 1/23 relates to chemical compositions, the decision has potential consequences for other technical fields. In this article, we consider implications for the seemingly distant field of computer-implemented inventions. The Decision In G 1/23, the Enlarged Board held that a product placed on the market before the filing date or priority date of a European patent application is part of the state of the art, even if its internal structure or composition cannot be analysed and reproduced by a skilled person. “A product put on the market before the date of filing of a European patent application cannot be excluded from the state of the art… for the sole reason that its composition or internal structure could not be analysed and reproduced by the skilled person before that date.” For chemical compositions, this marks a departure from the stricter interpretation of G 1/92, which had been read by some to require that a product must be both analysable, as well as being reproducible without undue burden, to qualify as prior art. G 1/23 clarifies that public availability alone is sufficient, and that technical features become part of the state of the art if they are accessible, even if not fully understood or replicable. Notably, while G 1/92 also concerned chemical compositions, the decision has been applied in other technical domains, including software (e.g. T 2440/12) and computer hardware (e.g. T 1169/04), suggesting that the reasoning in G 1/23 may similarly influence interpretation across a broader range of technologies. Implications for Software Software companies often deploy their products via the cloud (such as a Software as a Service or “SaaS”), offering services through web interfaces or APIs without disclosing the underlying source code or architecture. Historically, this raised questions: if a competitor couldn’t reverse-engineer the internal workings, could the software still be considered prior art? G 1/23 confirms the previous conclusion under G 1/92 as applied in T2440/12, in which the Board considered that if the skilled person had access to a copy of the software, then that is considered a disclosure of the underlying algorithms. In particular, the Board held that the public sale and use of a software product that implemented the claimed method rendered the method not novel, even though the source code or algorithm was not explicitly disclosed. This is because the software could be executed line-by-line on a virtual machine, and this execution would reveal the method’s operations to the user, and also amount to reproducing the method. Under G 1/23, the Board in T 2440/12 would not need to assess whether the skilled person could reproduce the method, but given that line-by-line execution was found to constitute both determining the operations and reproducing the method, the Board would almost certainly come to the same conclusion. The same reasoning would apply to most instances of locally-installed software, providing the user can run it on a virtual machine, even if encrypted or otherwise obfuscated to prevent reverse-engineering. Where a software product is publicly accessible through a SaaS platform, the “virtual machine” reasoning no longer applies, because the user does not typically have the same level of control over how the software is executed. It is, therefore, at least possible that G 1/23 would yield a materially different conclusion to G 1/92 in the context of SaaS. In particular, the removal of the reproducibility requirement could in principle lower the bar for what qualifies as being state of the art. In the absence of any Board of Appeal decisions addressing this point directly, a common sense reading of G 1/23 should be followed, which is that the state of the art contains whichever technical features implemented by the service are discernible during standard use or described in documentation, even if the skilled person cannot replicate exactly how the software works internally, regardless of whether the method could be reproduced without undue burden. Conclusions Our analysis shows that fears of G 1/23 introducing an on-sale bar in Europe are overblown. Although the decision could potentially lower the bar for SaaS software becoming part of the state of the art, any change is hypothetical and likely relatively minor. The reality is much closer to “business as usual”. Nevertheless, the case has given us an opportunity to re-examine the boundary of the state of the art for software. Some practical consequences are as follows: Patent applicants cannot rely entirely on the opacity of cloud infrastructure to preserve novelty of their system under the European Patent Convention. Any decision as to whether to file a patent or rely on trade secret protection for a SaaS product must account for the reality that once the product has been released or put into service, patenting may no longer be an option Examiners or third parties may cite publicly available software as prior art, even if they cannot replicate its internal mechanisms. How we can help EIP’s Codiphy team, comprising software-focused patent attorneys and commercial IP solicitors, has deep expertise in all aspects of IP law as applied to software technology, and can offer unparalleled strategic guidance for companies deploying SaaS or other software solutions. For example, we can provide: “Code-to-patent” mapping: helping clients identify which parts of their software are patentable and which are better protected as trade secrets. Disclosure risk analysis: assessing which aspects of a cloud-deployed product may be considered prior art following the decision in G 1/23. Filing strategy advice: advising on when and how to file to preserve novelty while aligning with product release timelines, especially in light of the jurisdictional differences in Europe and the US. Competitor intelligence: monitoring public-facing software and documentation that may now be used as prior art.