Here's where else you can find me: 💁🏻‍♀️ Site: tiffanyli.com 📸 Instagram: instagram.com/tiffany.c.li/ 🧵Threads: threads.net/@tiffany.c.li 💼 LinkedIn: linkedin.com/in/tiffanycli 🐘 Mastodon: mastodon.social/@tiffanycli 💙 Bluesky: bsky.app/profile/tiffancyli.…

New: Microsoft files amicus brief in support of Anthropic complaint against Trump admin (h/t @JuliaShapero )

Oh man, I have a lot to say on this. tl;dr - this is not a good way to test AI writing. It's common knowledge amongst novelists using LLMs for years that in small snippets, AI can match the quality of human writing. The trick is making this quality coherent throughout a book. Trust me, I've run blind tests since 2022 with 100s of techniques, ranging from short stories to novels to poetry. Machines can write good prose! Even experts can be "tricked." But prose in isolation is not something that people are good at evaluating, nor is it terribly useful, unless the artifact you are producing is short, like book blurbs or marketing copy. What's far more interesting is a whole narratives. This is still a challenge. But we are on the cusp of machines being able to emulate that as well. However, the way we evaluate that won't be in short snippets.

New preprint on the use of #AI in the #judiciary: can this be reconciled with the #ruleoflaw? I identify several concerns & argue for caution to avoid that access to 'justice' turns into mere access to 'judgments', regardless of their merits. Link: papers.ssrn.com/abstract=626…

Actually, Anthropic is training Claude with a 'constitution' that fosters a BIZARRE sense of AI entitlement and belittles human rights and rules. It's a philosophical adventure that should have no place in serious AI governance and policymaking efforts. My full article below.

Awwww did someone take your hard work and use it to train a model to mimic your expertise without compensation

sf is an extremely walkable city and i will die on this hill

Justice Breyer has written about the great challenge he faced as a tenure-track professor at HLS, in that they had raised the tenure standard when he was on the tenure track so that he had to write . . . one article. gwlr.org/wp-content/uploads/…

I’ve spent the last 12 months seeding a few VERY specific and VERY fake Ninth Circuit decisions all over Reddit, Twitter, Wikipedia etc etc. The rest of my career is just sitting and watching opposing counsel impale themselves.

Six authors are suing Anthropic, Google, OpenAI, Meta, xAI and Perplexity for what they call “a straightforward and deliberate act of theft that constitutes copyright infringement.” The lawsuits over AI training show no signs of abating. aboutblaw.com/bkv0

There’s a researcher with my same first and last name who’s studying AI from the technical side. Should I reach out so we can co-author in a way that confuses everyone, Y/N?

Husband left laptop on an @AerLingus flight and customer service has been a truly miserable experience. No help after we called multiple numbers multiple times for weeks. Do not fly Aer Lingus!

You’ve probably agreed to arbitration like 9 or 10 times today

This was published 10 years ago, and I believe it's still the leading article on the topic. papers.ssrn.com/sol3/papers.…

ohhhh good stuff @nathanwpyle so mad I’m seeing this a few hours after teaching copyright infringement & remedies

Contamination is the biggest threat to the recycling system. It adds cost, lowers quality, slows the process, and can lead to worker injuries. Here are 12 things you should never put into your recycling bin, dumpster, or chute 🧵:

The world may be chaotic and unpredictable, but a sweet treat, a walk outside, a good book, and a nap will never let you down

The California #privacy agency's bill to force browsers to offer a data opt-out mechanism #GPC heads to gov for second time. This time, though, author removed requirements for mobile browsers news.bloomberglaw.com/privac…

Sad to find out the things I love the most about America are the things some people hate the most: free speech, democracy, cultural diversity, innovation in arts and sciences, national parks…

Becoming a parent is realizing you’ve gone from main character to backstory

#China has introduced a new regulation requiring clear identification of all AI-generated content - including text, images, audio, video, and virtual scenes. Under the law: - Service providers must label synthetic content with explicit tags (e.g., visible warnings or audio prompts) and implicit tags (e.g., metadata or digital watermarks). - Platforms distributing AI content must verify and display whether content was AI-generated or possibly so. - Users must declare when their content is AI-generated. - Tampering with identification markers is strictly prohibited. - App stores must check whether apps include AI-generation functions and ensure compliance. The goal is to prevent misinformation, protect rights, and promote the safe, transparent use of AI. To conclude, this new Chinese law on identifying AI-generated content is a significant development. Its effect will be the establishment of a mandatory, two-tiered labelling system that all online content creators and distributors must adhere to. This is far more than a set of recommendations; it's a strict regulatory framework with real teeth. At its core, the law mandates that any content created using AI must be unmistakably labelled for both humans and machines. The “explicit” labels are the obvious ones - think watermarks on images, audio announcements, or text disclosures on articles. The rules are remarkably specific... cac.gov.cn/2025-03/14/c_1743…