In general, the main obstacle to obtaining patent protection for a new software invention is the need for that software invention to be viewed by the relevant patent office(s) as “patent eligible”, i.e. as the sort of invention for which patent protection can be obtained.
This blog seeks to provide some high-level guidance on what sorts of software invention are viewed as “patent eligible” by most major patent offices.
Generally, when it comes to determining patent eligibility of a new software process, the technical effect of that process is more important than the software technique used to achieve that effect. So, for example, most patent offices will readily consider new software which incorporates a novel AI or machine learning technique as patent eligible, provided that novel AI/machine learning technique provides some sort of useful technical effect in a non-obvious manner. In contrast, software which applies a novel AI/machine learning technique to a non-technical (e.g. business-related) problem will not normally be seen as patent eligible because there is no technical effect. Equally, software which applies a known AI/machine learning technique to a known technical problem to produce an expected technical result will not normally be seen as patent eligible, because this will be seen as an “obvious” thing to try.
Of course, there are plenty of grey areas where patent eligibility is less certain than in the above cases. For example, a new software process may fall in such a grey area if the software process:
For software inventions falling in grey areas such as these, jurisdictional differences tend to become more pronounced, and certainty is generally not possible. So, where there is doubt about whether new software is patent eligible, it is highly advisable to seek advice from a patent attorney with relevant software experience.