New language for our publisher guidelines? "We don't knowingly sell titles that employ generative AI in any capacity. If we find out a particular title was created using AI, we reserve the right to remove AI-generated portions or delete the title from our offerings entirely." Opinions welcome
I'd suggest striking "to remove AI-generated portions" and stick to simply deleting the title from your offerings entirely. As written, removing portions of a work puts the onus on you to determine what those are and how to excise them. That's complicated. IE, what if the author claims it wasn't AI?
And if it's a case where a story in a magazine is later revealed to be LLM-generated, the publisher of that story could always resubmit an updated file.
Excellent point. To be honest, I got kind of excited imagining a case in which we fulfill the expectations of our subscribers but still get to make our point by sending them an issue of a mag with a gray box in place of its formerly ai-generated cover :D
That sounds incredibly satisfying personally. But I feel like it might open you up to potential legal liabilities. Possibly I've been lawyer-adjacent for too long...
I am no lawyer! But in my limited defense, this indeed very satisfying fantasy was based on a bit from the USPTO guidance about copyright and AI generated work: www.federalregister.gov/d/2023-05321...
I think you're among anti-LLM allies, no need for defense here! My mind is just going straight to all of the spurious suits/rumor mills people can foment. I live in a depressing copyright world sometimes.
I've added a disclaimer to my website:
"Redfern does not use generative AI for any part of their writing at any stage."
I suggest other authors to the same. If you don't want to write, then don't be a writer!
I have a 4kw story that was the result of idly riffing off a 1-line post from a Twitter bot. According to everyone I've shown it to it slaps. I'm getting hassled to sub it.
It's about consequences of ubiquitous AI, but its unique launch point is that line.
I have no idea if it's even worth subbing
@interzone.press on the basis that you've closed subs for the foreseeable, this isn't a pitch - where do you draw the line on this sort of thing?
If it's all the author's original work, but their writing prompt was a line from a bot?
Less assisted by AI than burgled from AI, but a waste of time?
Weightless sells ebooks from other publishers, we don’t publish any ourselves, but IMO this is worth submitting. Taking inspiration from one line by a bot is like taking inspiration from an interesting mold pattern on a wall: still very human
The issue primarily is that aside the creative work it has stolen for training data, the main thing AI writing has stolen is time: the time and energy of editors.
I understand the importance of drawing a hard line against "AI assisted" submissions and am concerned that this is too much a grey area.
As someone who, for several years now, has been asked on a daily basis to "tidy up" AI output (in this case, translation), I feel this to the depths of my soul. You can't tidy it up. You can make it slightly less shit. But it never becomes good unless you retranslate it from scratch.
Indeed. As a translator you must be bombarded with it.
I'd never consider submitting something written by AI or even reworked from AI output for publication.
The only reason I'm on the edge about this one is that I wrote it. It's all my work, with the exception of the 11-word bot post I riffed off
Nah, a bot wrote some random words and you thought "Oh, I wonder what would happen if". That's just normal inspirational processes. So the story is totally your work, and you shouldn't feel remotely anxious about it.
