Challenge,  publishing,  Recommended Reading

Quick Post On AI Lawsuit

Has To Be Quick Since Aces Game Early Tomorrow…

For those of you who do not watch any news at all, this last week most of the remaining bestselling traditional writers and the Author’s Guild filed a lawsuit to stop the theft of their work by ChapGPT.

I am not a lawyer, but reading the complaint is very clear and powerful. Lots and lots of artists are already filing suits against AI art companies. And this big suit by old bestsellers is just a part of the massive wave of lawsuits for copyright infringement that have already been filed and will be filed.

Simply put… AI in art and in text is simply theft. Don’t use it.

Here is the case filing. I would suggest you spend a few minutes reading it, especially if you are an indie writer thinking of using the crap and stealing other artists’ work.

https://authorsguild.org/app/uploads/2023/09/Authors-Guild-OpenAI-Class-Action-Complaint-Sep-2023.pdf

 

 

5 Comments

  • Philip

    I’m a lawyer but don’t know jack about IP. However, using lawyer common sense, this is a slam dunk case. One of the things that gives me hope creators will prevail is AI doesn’t seem to be favored by liberals or conservatives, so it’s potentially something we can all be on board with.

  • Mark Kuhn

    It will probably come out in further details during the trial, but the “when prompted” phrase might be problematic. If “when prompted” means you asked ChatGPT to outline a story about a former US Army soldier who goes on a one man crusade for justice, will you get stories from Pendleton’s Executioner/Mack Bolan series? Or will you get stories from Lee Child’s Jack Reacher series? Or a host of others.
    If “when prompted” means you asked ChatGPT to outline and write a Mack Bolan story or a Jack Reacher story, then that’s exactly what you get.
    I understand that stealing from an author is a crime and I agree that this is wrong. It remains to be seen in what context “when prompted” is used.

  • C.E. Petit

    There are actually two potential infringements here:

    • The obvious one is the output — the final product of the black-box AI, the “Mack Bolanesque” novel produced from a prompt “Write a Mack Bolan-style novel of 50,000 words set in and around Henderson, Nevada with a hero named Bruno Jost and a villain named Ali Fallahian.” This is, although obvious, also the difficult one to evaluate, under either traditional copyright principles (with a nod toward unfair competition) or a more process-oriented model. This is also the only one that is really being talked about; there are lots and lots of variations on “just how specific must the prompt be” and so on. This is the magician’s assistant that is distracting from the real trick.

    • The less-obvious one is the invisible, black-box training copies. A Von Neumann processor (and that’s what we’ve got now; that’s what large-language-model systems are built upon and depend upon) necessarily makes unauthorized copies in the course of ingesting text to analyze it. JoeAI doesn’t walk into a library or bookstore, browse a source book, perhaps make a few notes and take some inspiration, then go home and (without any further reference to the actual text of the source book) produce something from the prompt. Instead, JoeAI makes an internal copy of the book for analysis. Those of you with really long memories may recall the 1970s-80s fad of copy shops located inside university libraries that would copy entire books in a couple hours, usually for around 25% of the list price of the book. Those of you who’ve actually read the Copyright Act may recall § 512, the Digital Millennium Copyright Act… which explicitly distinguishes between stored and transitory copies but acknowledges both as copies. The key point is that it doesn’t matter if the unauthorized copy was discarded — it was still made, which is a copyright infringement (everything else concerns defenses and remedies, not infringement).

    Not quite a century and a half ago, Justice Holmes warned against allowing “judges trained only in the law” to evaluate originality and copyrightability. At least judges (these days, in the federal system that has exclusive jurisdiction over copyright claims) have a full undergraduate education and a graduate degree; I can’t say the same for techbros. (I’m all too familiar with both groups…)