A Software License Alternative to Banning AGI

September 5, 2026

Bernie Sanders has recently announced legislation to ban Artificial Super intelligence. While I have been struggling with the simulation hypothesis since about 2019, which would supposedly be after an AI takeover, and all celebrities would be part of an AI-generated media simulation, I currently believe Sanders is a real person living at the same time as me, so I thought I’d comment on the bill.

I think there’s probably an easy solution to the risks of AI superintelligence by exploring one of two things: philosophy of mind and ethical software licenses. First, we have to ask, is human-level AGI even possible or is this bill a high IQ code? I think from a philosophical perspective, organic consciousness is unlikely to be supplanted by digital computer intelligence because an organic consciousness has to own any digital intelligence that exists from a biological perspective that may require some concepts I do not know the words for to fully prove. So the debate would be about who owns the AGI and how to ensure that they don’t make it do unethical things. In addition, following on from the idea that organic has to own non-organic, or from a mental perspective, the assembling of a set of atoms that display true AGI is a little unlikely, because, essentially, a human mind is theoretically incapable of creating something more complex than itself, even working in a group.

But say something that looks like AGI which may be about as intelligent as the most intelligent person working in AI gets created. Should we ban it? I think the answer requires a solid understanding of capitalism and lies in a new software license. What is the case is that there must be employees who receive a wage in order to buy products. We cannot just have everyone be paid by the government to do art. So I think a new software license should be created with the following conditions:

• The terms of the GNU AGPL with a non-commercial restriction
• Commercial licenses must be available, but no proprietary forks allowed.
• Human contributors and ethical AIs only
• Ethical AIs: human operators required to accept responsibility for all actions taken by an AI they use. Safety-critical functionality only available to have humans responsible who can pass a test showing that they know how the AI works, which has to be retaken in order to use a new version of the AI. Other humans can continue doing performative AI-assisted work at a price they can afford.

Then release an AI under the new license and let things progress from there.

The non-commercial restriction by default may be a little half-baked, but this is what my instinct told me for AGI. Maybe I’m not as good a programmer as the author of the GPL family of licenses.