Three Ideas for Regulating Generative AI

I agree that transparency is important, but I don't agree that transparency can be necessarily mandated.

In the case of open-source technologies, those are inherently transparent and this is great, but unless something about software is drastically changed, nothing is actually required to be open source, and due to this tradition it would require radical change in all software standards to mandate everything to be open source. Companies like Apple and Microsoft would likely not exist or would exist in a much different form.

My fear is that being overly scrupulous about mandates on open-source will drive every software project to be closed source, which is antithetical to the desire for greater transparency. Companies are entitled to their trade secrets, which is enforced by a great deal of legal tradition in the US.

Furthermore, the first amendment prohibits mandates about the design and algorithms involved in software development, as the 9th circuit court of appeals has upheld the status of software code being protected speech under the first amendment.

I think this article touches on some really good ideas, but ultimately I think mandating ethics and transparency is like mandating altruism and public disclosure of trade secrets. U.S. legal tradition doesn't really allow for that in the framework of business.

The best we can possibly do is regulate to define what outcomes are acceptable, rather than regulate design standards. Regulating design is really a non-starter .