- cross-posted to:
- aicopyright@lemm.ee
- cross-posted to:
- aicopyright@lemm.ee
‘Impossible’ to create AI tools like ChatGPT without copyrighted material, OpenAI says::Pressure grows on artificial intelligence firms over the content used to train their products
deleted
I am an attorney that has worked extensively in copyright and specifically fair use and the only interesting issue in copyright I feel like discussing is generative art — primarily because the discourse is getting overwhelmed by well meaning lefties and creators in my circles who are unfortunately carrying water for the same copyright maximalist arguments corporate IP holders have been marketing for years. We need more fair use, not less. Expanding copyright once again to cover training a model - which is what would be required because as is these models are not infringing on any existing copyright rights - would have disastrous results on small artists and creators, only surge to further enrich corporate IP hoarders that can handle the exploding court costs that will result from such an expansion.
To be clear, I also am very concerned about the many externalities of generative AI. Labor, environmental, competition, national security, privacy — but we’re not going to copyright our way out of those problems.
deleted
Yes, I have worked with small artists, creators, software engineers, game designers, YouTubers and podcasters. Among others. No, I won’t tell you who they are. No, I don’t care if you don’t believe me. These days it’s primarily privacy, to be fair, but yes, generative art and its intersection with copyright law is something I am passionate about. This isn’t the gotcha you think it is.
You write license agreements for your business partners. For most infringement you send a series of letters or try to interface with whatever cyberpunk dystopia bureaucratic machine is set up to handle DMCA with your platform of choice. Scraping otoh is something I advise to ignore. For those that are upset by scraping I would suggest that they can certainly use robots.txt to prevent future scraping but would also explain that their content isn’t getting stolen by OpenAI. What we’re seeing with generative AI is not an abuse of existing copyright, it’s something new, and I fundamentally don’t think more expanding copyright again is going to help.
Look, I’m no fan of Sam Altman. Fuck that guy. Fuck every billionaire. His crypto disaster is a privacy dystopic wet dream. I imagine you and I would probably agree a lot more than we disagree. The problem, as I see it, is that copyright has ALWAYS been a shit framework for small artists and creators. It’s an effective instrument to be abused by those with bottomless pockets. But it is fundamentally not suited to actually address the negative externalities we are all seeing with generative AI.
deleted
Not legal advice not your lawyer etc etc. But I would likely never suggest someone pursue aggressively against individual piracy. You write contracts for your partners. You fight businesses when they breach. You make great work and price it appropriately. You make your wins there and you do everything you can to not find yourself in a courtroom or arbitration if you can avoid it. You’re not winning any friends and you’re not saving yourself any trouble by raging against torrents. Especially for small creators the calculus never (imo) works out in their favor. More often than not, small artists and creators need to be much more concerned about and need help with being able to defend themselves against spurious accusations of infringement by larger corporate Ip rent seekers and more-or-less automated systems (again: cyberpunk dystopia).
Speaking personally I find the equivocation of “copyright infringement” and “theft” ridiculous. One download = \ = one “stolen” sale, and it never has. Theft requires depriving the original of the property, being able to exercise exclusive control over it. Conceptually it has always broken down when talking about digital goods.
deleted
I’m Incredibly worried about AI deepfakes and voice cloning for a whole host of reasons. It’s one of the things I think we are collectively least prepared to deal with. The privacy concerns, national security, cyber security - to say nothing of disinformation and yeah, labor impacts — we are fucked and not at all ready for this.
Name and likeness rights, rights of publicity though and privacy rights don’t stem from copyright and don’t require an expansion of copyright to further protect. There’s case law already preventing a business from cloning someone without their permission, and everyone will be paying very close attention to those parts of contracts moving forward, I’d wager. As to wholesale replacing actors and talent with generated content — yeah, I’m very worried a lot of artists and creative people are as fucked as the lawyers and the accountants and writers and everyone else when it comes to job displacement.
Again, despite your really aggressive tone, I’m telling you: we almost certainly agree more than we disagree. It is ghoulish watching studios rush to replace extras and voice actors and resurrect dead actors. True cyberpunk dystopia necromancer shit. I’m hoping that we see more victories won in this genuinely encouraging resurgence of labor (todays SAG AFTRA deal notwithstanding) and legislation directly addressing the labor impacts of AI more broadly. Different kinds of guard rails and safety nets. I just don’t think copyright is the answer you think it is to the horrors that we both agree are coming.