Finally finished the class... and the weekend is gone.
Follow
Click to Follow technollama
Andres Guadamuz
@technollama
Law and technology academic. Tweets about copyright, AI, NFTs, smart contracts, internet regulation, llamas, pandas, and cats. "Top End Techtastic". 
Andres Guadamuz’s Tweets
You can find some quality advertising here at Twitter dot com.
1
Well, the En Ef Tee bots are back in force. This site is going to the dogs.
2
Only reading headlines and news stories, you would be forgiven if up until now you had thought SBF was a teenager still driving around on a learner’s permit, who picked up crypto trading to avoid high school summer jobs that might force him to go outside.
49
123
764
Show this thread
It's interesting to note that v2.0 doesn't produce any watermarks.
2
9
Show this thread
Having some fun stuff experimenting for the class. Here's the same prompt in Stable Diffusion, DALL-E, and Midjourney. Only SD reproduces the Getty watermark. Prompt: "Old photograph of a busy street Getty Images".
6
4
21
Show this thread
Working on my AI copyright infringement class, I've had to completely rebuild it from scratch. Last year: "it's all theoretical". This year: "these are all the people getting sued".
4
33
162
Show this thread
Interesting example of a law class being ChatGPT friendly.
Quote Tweet
A l' @UNamur, on est #chatgpt FRIENDLY On a parlé de nos prémisses pédagogiques avec quelques collègues pour @lalibrebe Le meilleur reste à venir 
lalibre.be/etudiant/etudeShow this thread
2
3
Weekend?
read image description
ALT
1
7
Google announces Dreamix: a model that generates videos when given:
- video + prompt (Video editing)
- input images + prompt (Subject Driven Generation)
- input image + prompt (Image-toVideo
0:48
388.7K views
87
1,788
7,827
No user is going to go to DreamStudio, ask for an image with a Getty Images logo, and think that it comes from Getty, that's why they´re arguing dilution and tarnishment. I'm not convinced it is dilution, but again, I'm not an expert.
4
4
Show this thread
I'm not a trademark expert, but these arguments seem to me to be much weaker than the copyright ones. I'll be curious to read what trade mark experts have to say about this.
2
3
Show this thread
Trademarks protect reputation. Getty are arguing that Stability has used their mark in commerce, and that the resulting use "dilutes the quality of the Getty Images Marks by blurring or tarnishment", and these actions harm their brand and leads to dilution.
1
3
Show this thread
The trademark question will be interesting, it's easy to get a return that has the Getty Images logo, I got this with the prompt "Old picture outside the cinema Getty images". But logo reproduction isn't necessarily trademark infringement, ™ isn't Ⓒ.
1
3
Show this thread
Who knows if StabilityAI has the stomach to go all the way? If they do and lose, there could be appeals, so this will not be solved immediately, unless it's settled out of court.
1
3
Show this thread
The case will likely rest on the infringement claim, and the defendants are likely to argue fair use. Could go either way, and says that there are not many fair use cases in the District of Delaware. So nobody knows.
Quote Tweet
The District of Delaware hears *very* few fair use cases. That said, Judge Stark is currently presiding over Thomson Reuters v. Ross Intelligence. Ross filed a summary judgment motion last month claiming it was fair use to copy chunks of Westlaw to create a rival legal database. twitter.com/copyrightlatel…
1
7
Show this thread
The key claim is that Stability used LAION's links to make copies of works to train its model, these contain Getty images, and "at times it produces images that are highly similar and derivative" of its library. This is a much more accurate description than the class action suit.
1
3
Show this thread
The complaint is technically more accurate than the class action lawsuit. It consists of straight copyright infringement claim, a rights management information manipulation and removal claim, and a trademark claim.
1
1
4
Show this thread
A very strong complaint. First thing to note, LAION is mentioned several times, but they're not a party and not getting sued. I still think LAION is untouchable, it's very indicative that Getty doesn't want to touch it.
1
5
Show this thread
Turns out the English threat was just a diversion (unless they run two separate lawsuits in different jurisdictions).
Quote Tweet
BREAKING: Getty Images just filed a copyright and trademark infringement lawsuit against Stability AI in Delaware District Court. Getty alleges that Stability copied more than 12 million Getty photos to train Stable Diffusion. Full complaint here: copyrightlately.com/pdfviewer/gett
Show this thread
2
4
18
Show this thread
I made the mistake of falling down a Twitter rabbit hole where people talk of AGI in quasi-religious terms.
3
1
12
#FundedPhD opportunities alert! Do you want to contribute to the just transition to socially just, trustworthy, responsible, open, inclusive and sustainable AI? Come do a PhD with us at ! Click on 'Just AI Lab' for more information stir.ac.uk/research/resea #JustAI #LawPhD
1
12
16
I can't help but think that Musk is turning-off Twitter like David Bowman shut down HAL-9000 in 2001, bit by bit.
2
10
Microsoft to integrate ChatGPT into Teams. Please make Clippy the ChatGPT avatar.
4
13
27
Show this thread
Taught smart contracts, e-commerce implications of web3, and AI copyright, all in the same day.
I need a hug.
6
1
52
Mark Lemley, one of the country's leading scholars in IP and internet law, will represent in the closely watched copyright suit from a group of artists.
He's part of Lex Lumina PLLC, a recently formed firm composed of a pretty formidable group of IP scholars.
4
23
43
Show this thread
Netflix accidentally posts guidelines for cracking down on password sharing. They were being trialled in Chile, Peru and Costa Rica.
AI, we come in peace.
Here’s our letter today in the .
, , , FYI 😉
27
71
225
It´s interesting that nobody is talking about the other recommendation, namely leaving the authorship regime intact. Not for the first time I note that the authorship question is becoming a bit of an academic discussion, the interest is in infringement.
1
1
4
Show this thread
If there's a consultation however it should be swift. As things stand the existing law is the 2014 exception for research, which opens the door for datawashing. Reform is needed and should still be swift.
1
1
5
Show this thread
Pretty big development, ministers commit to not go forward with UKIPO suggested change to text and data mining exceptions, and promise further consultations. I'm on record that the changes should match the EU's approach from the DSM Directive allowing for opt/outs.
Quote Tweet
1/ The Government's 'text and data mining' exception to UK copyright law would have left those in creative industries without legal control over their work, should it have been used by an AI platform.
I therefore held a debate today to highlight the importance of this issue.
Show this thread
1:38
38.9K views
4
6
23
Show this thread
I'm perennially amazed that Astrology is still a thing.
8
1
24
great paper, but what an odd summary
of 175.000.000 generated images, they managed to recreate 109 "near copies" of pics in the training set, that's 0,000062%?
alt headline: it's lottery-level difficult (but not impossible) to recreate images from the training data in SD 🤷♂️ twitter.com/Eric_Wallace_/
This Tweet is unavailable.
25
57
443
This is a very interesting look at the generative AI market landscape. While I tend to take anything VCs say with a pinch of salt, the analysis appears to be quite accurate about how things stand as of today.
1
3
22
Talking about anything related to internet regulation increasingly feels like an exercise in ancient storytelling.
"Gather around children, and let me tell you about this thing called Napster..."
4
10
48
This is my 100th 3.
Wordle 592 3/6*
⬛⬛⬛⬛🟨
🟩🟨⬛⬛🟩
🟩🟩🟩🟩🟩
1
For the record, I'm available as an expert witness, and also for birthdays, weddings and bar mitzvahs.
2
10
Show this thread