Artificial intelligence is an extremely complex field of study and invention, but one which is seeing a significant increase in commercial market share. As more tech innovators succeed in bringing their AI products to market, the issue of intellectual property also becomes more prescient. To what extent can a business ‘own’ the source code for a machine learning algorithm – and how might it be protected?
Trade Secrets, Explained
To understand your situation and options as a business, you must first understand what constitutes a ‘trade secret’. Before the late 2010s, the term ‘trade secret’ was essentially a colloquial one, used to describe an invention, process or intellectual property with intrinsic value to a business – and which could existentially harm said business if revealed.
Trade secrets are a large part of the promotional work done by popular brands over the last 100 years. Fast food giant KFC has kept its signature spice blend a secret for generations. At the same time, the original recipe for Coca-Cola remains performatively locked behind a vault at the company’s Atlanta museum.
But the trade secret also performs a vital function for other businesses less concerned with PR; keeping certain designs or inventions secret enables them to maintain their competition against rival companies, and to improve their products unilaterally.
Patents and Trade Secrets
As indicated before, the concept of the trade secret held little water legally, besides in relation to specific readings of intellectual copyright law. Instead, businesses seeking to protect mechanisms or inventions actively would use the patenting process – which formally recognizes a design or invention, and assigns exclusive usage rights for a set period of time.
Today, though, the trade secret has been enshrined in new legislation. The Trade Secrets Regulations 2018 protect bespoke inventions and ideas from illegal dissemination, where it can be reliably proven that such secrets are instrumentally valuable to a business.
Trade Secrets and AI
The field of AI is already something of a legal minefield, as the continued development of autonomous, self-learning technology raises new questions about liability and legal launch. With specific regard to trade secret legislation, AI is also unique; for many developers and innovators, the inner workings of advanced machine learning systems are an inscrutable ‘black box’ – impossible to understand or reverse-engineer.
As such, neither trade secret legislation nor patent technology can properly apply to the specific iteration of an AI or machine learning system. Instead, it would be the developer’s approach to generating such an AI that would be subject to such protections.
Follow Technoroll for more!
Editorial Staff of the TechnoRoll, are a bunch of Tech Writers, who are writing on the trending topics related to technology news and gadgets reviews.