A July 24 open letter from 26 industry heads urges AI leaders to expand access and make public their AI models.
To be sure, as the letter makes clear, opening up these AI models to the public presents a multi-edged sword, including an unusual set of opportunities for collaboration, leveling of the playing field for smaller companies and entrepreneurs and exponential growth; on the other hand, it also presents the potential for great danger and risk for the corruption of these AI models by bad actors.
On a far smaller scale, though, it raises an interesting question as to whether AI trade secret theft claims like those most recently brought by Apple against OpenAI would survive, and if so, what those claims might look like in an Open AI model world.
The July 24 Letter
Using a broad comparison to the explosion of open source software and the vast societal benefits it conferred upon the public, these the heads of 26 industry leaders, including Google, Microsoft, Meta, Palantir, IBM, Cisco and NVidia made the following argument:
Open weights expand access to the AI economy. Startups, established businesses, universities, and public institutions can build on advanced models without training one from scratch or paying frontier-model prices for every task. Open weights let every organization match the right model to the right job at the right cost, reserving frontier-scale capability for genuine frontier problems and running efficient, specialized models everywhere else ...
Open weights also give customers greater control. As organizations invest in AI, they want to know that they will not become locked into a single provider or lose the knowledge and capabilities they build over time ...
To be sure, open weights carry real and distinct risks. Once released, the weights are beyond the original developer’s control, and modified versions are difficult to trace or reverse. But the right response to this risk is not to prohibit open weights. In a world where cybersecurity attackers use advanced AI, defenders need access to models with comparable capabilities so they can detect, simulate, and respond to emerging threats.
If Adopted, Would Open Weight Models End AI-Related Trade Secret Theft Claims?
Without question, a cynical view of this letter finds plenty of grist for the mill, especially when you consider that one of the principal signatories of this letter is none other than OpenAI - yes, the very same OpenAI who is currently in Apple's legal crosshairs in the Northern District of California. After all, the argument goes, if all of this code and modeling is open to the public, that inherently undermines any claim that it is worthy of trade secret protection.
On the other hand, a closer reading of Apple's July 10 complaint provides a roadmap for what future claims of this nature will look like - which hearkens back to the gravamen of traditional faithless servant and misappropriation claims - causes of action that do not inherently rely on a "trade secret" designation in order to survive. Rather, these claims would focus on proprietary and confidential company material that is shielded from the public domain, and then downloaded (or forwarded) to departing employees' personal devices, email or cloud accounts when on their way out the proverbial door.
Therefore, on the surface of things, the suggested proliferation of open weight AI models shouldn't have an outsized impact on trade secret litigation surrounding AI model-related theft. That said, there is also no question that the optics of pursuing such a claim would be significantly more difficult, as the company would have to overcome the hurdle of demonstrating that the court should step in to protect something whose legal status as "proprietary" or "confidential" may be highly suspect.
