AI legal links

part 1: copyright

Generative AI’s Illusory Case for Fair Use by Jacqueline Charlesworth :: SSRN The exploitation of copied works for their intrinsic expressive value sharply distinguishes AI copying from that at issue in the technological fair use cases relied upon by AI’s fair use advocates. In these earlier cases, the determination of fair use turned on the fact that the alleged infringer was not seeking to capitalize on expressive content-exactly the opposite of generative AI.

Urheberrecht und Training generativer KI-Modelle - technologische und juristische Grundlagen by Tim W. Dornis, Sebastian Stober :: SSRN Even if AI training occurs outside Europe, developers cannot fully avoid European copyright laws. If works are replicated inside an AI model, making the model available in Europe could infringe the right of making available under Article 3 of the InfoSoc Directive. (while the US tech industry plays with the IT equivalent of shoplifting comic books, the EU has grown-up problems to worry about.)

Case Tracker: Artificial Intelligence, Copyrights and Class Actions is a useful page maintained by attorneys at Baker & Hostetler LLP. Good for keeping track of what’s where in the court system.

Copyright lawsuits pose a serious threat to generative AI The core question in fair use analysis is whether a new product acts as a substitute for the product being copied, or whether it transforms the old product into something new and distinctive. In the Google Books case, for example, the courts had no trouble finding that a book search engine was a new, transformative product that didn’t in any way compete with the books it was indexing. Google wasn’t making new books. Stable Diffusion is creating new images. And while Google could guarantee that its search engine would never display more than three lines of text from any page in a book. Stability AI can’t make a similar promise. To the contrary, we know that Stable Diffusion occasionally generates near-perfect copies of images from its training data.

When does generative AI qualify for fair use? Because fair use is determined on a case-by-case basis, no broad statement can be made about when generative AI qualifies for fair use. Instead, I’ll provide a specific analysis for ChatGPT’s use of its training data, but the same basic template will also apply for many other generative AI products.

The Heart of the Matter: Copyright, AI Training, and LLMs by Daniel J. Gervais, Noam Shemtov, Haralambos Marmanis, Catherine Zaller Rowland :: SSRN

Every AI Copyright Lawsuit in the US, Visualized | WIRED

Meta mocked for raising “Bob Dylan defense” of torrenting in AI copyright fight

part 2: defamation

KI-Chat macht Tübinger Journalisten zum Kinderschänder - SWR Aktuell

OpenAI, ChatGPT facing defamation case in Gwinnett County Georgia | 11alive.com

part 3: antitrust

Hausfeld files globally significant antitrust class action against Google for abusive use of digital media content Publishers have no economically viable or practical way to stop [Google Search Generative Experience] SGE from plagiarizing their content and siphoning away referral traffic and ad revenue. SGE uses the same web crawler as Google’s general search service: GoogleBot. This means the only way to block SGE from plagiarizing content is to block GoogleBot completely—and disappear from Google Search.

The Case for Vigilance in AI Markets - ProMarket (competition regulators in the USA, EU, and UK are getting involved)

part 4: false advertising

Google pulls AI Gemini demo video after National Advertising Division complaint | Ad Age The tech giant was not forced to delist the video, but voluntarily chose to do so in agreement with [The National Advertising Division (NAD) of non-profit BBB National Programs]

part 5: privacy

Meta AI Keeps Telling Strangers It Owns My Phone Number - Business Insider

Do LLMs Contain Personal Information? California AB 1008 Highlights Evolving, Complex Techno-Legal Debate (California law requires opt-out and right to delete for personal data incorporated into an ML model.)

part 6: misc

AB 2013: New California AI Law Mandates Disclosure of GenAI Training Data | Perkins Coie AB 2013 mandates that the developers of GenAI systems or services post detailed documentation on their websites about the data used to train these systems. Required disclosures include Whether the datasets included any data protected by copyright, trademark, or patent or are entirely in the public domain, and whether the datasets were purchased or licensed by the developer.

AI models are being blocked from fresh data — except the trash – Pivot to AI We knew LLMs were running out of data as they had indexed pretty much the entire public Web and they still sucked. But increasingly AI company crawlers are being blocked from collecting more — especially data of any quality

NaNoWriMo Shits The Bed On Artificial Intelligence (imho they’ll figure this out before November, either the old org will reform or a new one will launch. Recording artist POVs on Napster were varied, writer POVs on generative AI, not so much.)

Is AI a Silver Bullet? — Ian Cooper - Staccato Signals TDD becomes a powerful tool when you ask the AI to implement code for your tests (TDD is already a powerful tool, and LLMs could be a good force multiplier. Not just writing code that you can filter the bullshit out of by adding tests, but also by suggesting tests that your code should be able to pass. If the LLM outputs a test that obviously shouldn’t pass but does, then you can fix your code sooner. If I had to guess I would say that programming language advocacy scenes are going to figure out the licensing for training sets first. If the coding assistant in the IDE can train on zillions of lines of a certain language because of a programmer co-op agreement, that’s an advantage for the language.)

Why A.I. Isn’t Going to Make Art

Have we stopped to think about what LLMs actually model? Big corporations like Meta and Google tend to exaggerate and make misleading claims that do not stand up to scrutiny. Obviously, as a cognitive scientist who has the expertise and understanding of human language, it’s disheartening to see a lot of these claims made without proper evidence to back them up. But they also have downstream impacts in various domains. If you start treating these massive complex engineering systems as language understanding machines, it has implications in how policymakers and regulators think about them.

Slop is Good Search engines you can’t trust because they are cesspools of slop is hard to imagine. But that end feels inevitable at this point. We will need a new web. (I tend to agree with this. Search engine company management tends to be so ideologically committed to busting the search quality raters union, and other labor organizing by indirect employees, or TVCs, that they will destroy the value of the search engine to do it.)