If asked a certain question, chatgpt quotes me and only me from a made up Wikipedia entry that I wrote in a hot minute over a bet with girlfriend circa 2007.<p>Wikipedia never deleted it for a couple years and even added to it because well, people.<p>It was written as a complete falsehood so I could win a 25 dollar bet with my current girlfriend.<p>As such it is an original work of fiction.<p>Since chatgpt is a for profit endeavor, I am going to assert my common law copyright against open AI and Microsoft until they rewrite the response to not reflect my and only my exact words as if it were fact.<p>To those of you whose code and art has been scanned for training I feel sorry for you.
Rights have been abrogated in the past. AI will change the way we treat original creations. Consider the following historical example:<p>>The maxim "Cuius est solum, eius est usque ad coelum et ad inferos," which is Latin for "Whoever owns the land, owns everything up to the sky and down to the depths of the earth" was used to describe that a property owner has complete control over the land they own, including everything above and below it.<p>...<p>>In the United States, the Supreme Court case United States v. Causby (1946) addressed the issue of airspace rights and established the concept of "navigable airspace." The court held that property owners have exclusive control over the airspace above their land up to a "reasonable" altitude that is necessary for the operation of aircraft.<p>>Since then, other cases and laws have further defined the limits of property owners' rights to airspace, including the Federal Aviation Act of 1958 and subsequent amendments. These laws and court decisions have essentially set aside the idea that property owners have complete control over the airspace above their land in order to accommodate the needs of air travel.
I understand where you are coming from and agree that this is a massive abuse of authorship rights, but I do not think it should be required to go through this complex process.<p>We know the network only works thanks to training data. We know training data was scraped without consent. We know the output is derivative of training data. We know OAI/MS get paid for those derivative works and are not in turn paying the producers of the original content. We know it has real impact on those producers and will cost future incentives to publicly share original research, along with other societal implications.<p>I wonder when large publishers in the US and EU wake up to the fact that their books and films were also scraped the very same way. Unless MS is paying <i>them</i> behind the scenes, they have a corporation profiting off their publications and they are not known to be especially charitable.
I don't know (US?) Copyright law in any detail, or any copyright law beyond reading debates about GPL and other licences (IANAL).<p>but wouldn't the fact that you have successfully passed off this fiction as fact for 16 years count against you in court? In a related point, do you have to assert copyright (like trademarks) to retain your rights? Have you done this with other users of your work? If not, then your failure to assert copyright previously may count against you.
The idea that training a machine on fiction owned by others and facts published with distribution rights can be surfaced in a LLM is not only absurd, it's illegal, locally and internationally.<p>Bring your horse and your carriage, you will lose to the human race before your horse clears the first turn.<p>ChatGPT is Sam Altman's "last scam"
Data was open to anyone all the way over the internet, the science of doing something great and also the hardware for blending all together that is what’s most valuable. Must be some big picture that driven all this new chapter of generative stuff.