Trade Secrets Fill The Gap Left by Patents in AI Protection

In March 2020, Neural Magic sued Meta and a former employee, for misappropriation of trade secrets.

Neural Magic accused Meta of stealing machine-learning algorithms via Dr. Aleksandar Zlateski, an ex-employee who left to join Meta.

The matter revolved around Meta’s publication of a software programme that contained the same proprietary algorithms developed by Neural Magic.

The court ruled that the plaintiff was allowed to offer testimony from an expert who said Meta owed as much as $766 million in royalties.

The parties settled shortly before trial!

Cases involving the misappropriation of trade secret technology can often involve significant damages.

Patent protection can be important, but AI innovations move fast, and patents take years also requiring public disclosure.

Many AI advances (e.g., tweaks to models, data labelling methods, or deployment tricks) are not easily patentable but are highly valuable if kept trade secret.

Just how are you protecting your AI competitive edge from mobile talent pools?