There has been much talk about AI, Artificial Intelligence, and its present and future effect. More than a few think it will be the destruction of humanity. I don’t go that far, but do recognize some of the problems. Here is a way to mitigate many of them. I don’t know how to write legal documents so don’t even try. Here is a starter list of the concepts. Please remember, it is just a starter list and needs significant work.
Basic Requirements
Every AI application must annotate everything it creates or modifies as an AI product. This means text documents, pictures, videos, and audio. And certainly, any other AI product not mentioned here. It must also be illegal to removed the AI notice from any such products.
Failure to annotate as required, or to remove such annotation, must be punishable at some severe level such as a felony conviction and fines that can be quite substantial. Those penalties, to varying degrees, should be applicable to not only the creators, but the owners of the app and the owners of the computers where the app is running. AI software and hardware not compliant must be shut down as soon as discovered. And yes, to include anyone presenting AI content to another person without disclosing the AI component.
Text documents
Each AI generated or modified text document must have a notice at the beginning that it was AI generated and an indicator as to which problem. In some place in the document it must show which AI program generated the document, when it was generated, and on what computer it was running. Complete identification must be required for all AI products.
Video
Each AI video or picture must contain an indicator that it was generated or modified by AI. Maybe something like a QR code that links to the description of the computers that generated or modified the document. The size must be sufficient for the average video device, cell phone, to view it and be able to follow the link. The indicator must be obvious to all viewers. It must be accompanied by some text indicating it was AI generated. It must be displayed at all times.
Audio
Every audio product, and every AI conversation must contain a notification that the audio or conversation was AI generated. Not certain how to do this, but it might be required to audibly simply state it was AI generated. When engaging in a conversation, every so often, maybe ever minute or so, it must remind the listener that the audio is AI generated.
Maybe the audio should have some type of indicator that, say, a cell phone, can detect, show some details, and provide a link to the system that has, or is, generating the audio. This part might not be discernable to a human, but, software examining a recording of the audio must be able to detect it. Engineers can easily find a way to implement this.
Software
This is for AI generated software. Depending upon how it is used, an AI marker must be present and easy to find. AI engineers and all creators are required to find a way to implement this marker. All programs created by an AI must be marked as such so then people working with it can not only easily discover the fact, but it must be obvious to the slightly knowledgeable user.
Conclusion
The fundamental concept is that to a significant degree, absolutely everything that is AI generated must be marked as such. Knowledgeable people can do a better job than I, but here is starter, a concept.