The AI & Legal Update #17 - 🎨 AI art and 95% of lawyers agreeing

Hi friends! OpenAI broke the internet with Ghibli-style images, NYT scored a big AI copyright win, and 95% of lawyers are bullish on AI. Let’s dive in.

📘 Read time: 8 minutes 27 seconds

Your editor with GC AI customers at the WGCN summit in DC last month ❤️

Big Tech 

🎨 OpenAI dropped a new image-generation model. Founder & CEO Sam Altman encouraged everyone to create images in the style of award-winning animated movie maker Studio Ghibli, which led to “biblical demand” as people made  AI-generated images like mine below. Also, a meme this week? Action-figuring yourself. Amazing.

This week, OpenAI also dropped that 1/10th of the world, so ~800 million people, use ChatGPT. Do you? Bonus points if you use GC AI and ChatGPT.

🏢 Google’s Waymo autonomous vehicle division announced its next city is… Washington DC. Regulators (those who still have their jobs, anyway 😢) will notice. Fun fact: DC has more lawyers per capita than any city.

Related: A security researcher uncovered a draft privacy policy suggesting Waymo One may use interior camera footage of riders to train AI models. Waymo has since clarified it has "no plans“ to use in-car video data for targeted ads.

But what about … privileged conversations? See GC AI analysis of whether privilege attaches to conversations in taxis (generally: no). 🙈

🐦 xAI, Elon’s AI company, bought X (formerly Twitter) for $33B in an all-stock deal. The stated plan is for Twitter to be a distribution mechanic for xAI. Can Musk just use all of X’s data to train its AI model? We asked GC AI. The short answer is: not so fast! See analysis/summary linked here and below. Related: Canada's privacy watchdog is investigating whether X's use of personal data for AI training violates privacy regulations.

AI & Regulatory

📄 The always-fascinating Pew Center put out a survey, "How the U.S. Public and AI Experts View AI.” Key findings: 

• 58% of US adults and 56% of AI experts say they’re concerned the US government will not go far enough to regulate the use of AI.

• 62% of U.S. adults and 53% of AI experts said they have little to no confidence in the U.S. regulating AI effectively (!)

🌐 The OECD rolled out a 29-criterion framework for reporting AI incidents. Check out GC AI’s guidance on how to comply - linked here and below. 

🇺🇸 US law firm White & Case put out a handy global AI regulatory tracker but did not also find the time to sign the amicus brief in support of law firm Perkins Coie’s challenge to the administration’s actions against the firm. Many reasons pro/con but biglaw friends reading this, your GC clients indeed notice these things.  

🏛️ The White House has ordered federal agencies to name chief AI officers. The new directive requires agencies to develop AI expansion strategies within six months, emphasizing a "forward-leaning and pro-innovation approach."

AI & Legal

📜 Judges on the Court of Appeals for the D.C. Circuit explicitly acknowledged in this opinion using ChatGPT for the drafting process, a big shift toward transparency about AI use in judicial writing. Meanwhile, some jurisdictions (looking at you, North Carolina! 👀) still require attorneys to certify they haven't used AI. The future happens in spurts …

On that note, the maker of Westlaw, Thomson Reuters, polled lawyers and found that 95% of them say AI will be central to their workflows within 5 years, while noting that the “[T]hree years [since ChatGPT’s release] is not a long enough time for many practitioners to have developed a concrete opinion.”

Tbh? We like that they did a poll, and it’s a good stat, but here at GC AI, we think AI being indispensable to legal work is happening sooner than that for the best in-house teams. Quote of the week, from a user who’d been on a GC AI trial that expired while we worked with her procurement team: “Being without GC AI the last three days has been like working without my right arm.” (!) Amazing. 💪

🧑‍⚖️ The American Bar Association released new AI Guidelines for courts. The recommendations focus on transparency, bias mitigation, and human oversight for court use of AI. See the excerpt on the use of GenAI below.

🧑‍⚖️ A Judge in Florida allowed a VR simulation of an alleged crime to be submitted as evidence. At issue was whether the defendant was properly invoking Florida’s stand-your-ground law. The defense attorney explained: “We put headsets on the judge, the prosecutors, and the witness, and the judge was able to see from my client’s own eyes, from his own perspective, what he faced when he was surrounded by intoxicated partygoers.”

AI & Corporate Adoption 

🤝Mid-wage professionals are adopting AI faster than their higher-paid counterparts. Why? Top earners fear looking replaceable or worse, like they’re cutting corners. But! As we all already know, lawyers who know AI actually make more money (see chart below). PS - Have you taken one of our prompting classes yet? If not, what’s stopping you? Bring on the future. 

VC firm Bessemer put out the Healthcare AI Adoption Index. According to the firm, Healthcare has reached its AI inflection point with 95% of surveyed healthcare leaders believing GenAI will be transformative, though only 30% of AI proof-of-concept projects currently make it to production despite significant investment. Turns out AI is a thing! Who knew? 😋 

AI Globally

🇪🇸 Spain approved a bill adopting guidelines from the European Union's landmark AI Act that fines companies up to €38M for failing to label AI-generated content - one of the strictest EU implementations yet. 

🚫 The European Commission released guidelines detailing eight prohibited AI practices. Fines of up to €35M or 7% of your global turnover. See GC AI summary here.

AI & Intellectual Property

The NY Times scored a big win against OpenAI last month when Judge Sidney H. Stein of the U.S. District Court for the Southern District of New York denied OpenAI's motion to dismiss NYT’s core copyright claims. The fair use question - can you train a model on publicly available data without infringing copyright? - remains open. OpenAI says they "look forward to making it clear that we build our AI models using publicly available data, in a manner grounded in fair use, and supportive of innovation." Notably, the judge just issued an order - he did not issue an opinion explaining, saying instead he will do so expeditiously. It’s like the Super Bowl is coming for copyright lawyers! GC AI analysis linkedin here and below.

©️The DC Circuit has ruled that AI systems cannot qualify as "authors" under copyright law. In Thaler v. Perlmutter, the court affirmed that works created "without any human creative contribution" cannot receive copyright protection. See GC AI summary here.

⚖️ OpenAI’s latest pitch? AI models should be allowed to train on copyrighted material without permission, warning that the US will lose to China otherwise. 👀

🇫🇷 France’s publishing groups are suing Meta for allegedly training AI on copyrighted works without consent.

🤔 Twitter/X founder Jack Dorsey & Elon Musk want to "delete all IP law", presumably so that content can be used for AI training. A curious take given ... all of software and social media is protected by the combination of IP and contract law? Weird. 

New, Notable & Random

💔 A journalist uploaded her chat history with her boyfriend to ChatGPT, which analyzed their relationship and found that he asks double the questions she does and told her to take more initiative. Hm.

⚡️“Thunderforge” is the Department of Defense’s flagship AI program through which the DoD will work with Anduril, Microsoft, Scale AI, and others to develop and deploy AI agents. “Thunderforge marks a decisive shift toward AI-powered, data-driven warfare, ensuring that U.S. forces can anticipate and respond to threats with speed and precision,” says the DoD’s Defense Intelligence Unit. 

✈️ Your editor is weirdly into airplane wing shots! Fun thing is giving ChatGPT a shot and asking it to geoguess the location. Notice ChatGPT is chatty and compliments my photography? Flattered! GC AI is more profesh tho.   

GC AI - Product Updates & Upcoming Classes

💪 GC AI’s core reasoner is on Anthropic Claude Sonnet 3.7 now. It’s more powerful and thoughtful.  

🆕 New prompt - approval cover email. Great for when you need to send an email to a stakeholder about a document they are signing. Example of this prompt here (reviewing this contract), and an in-depth explanation of the prompt and more examples here.  

👩🏻‍🏫 AI Prompting for In-House Legal - 101 on Thursday, May 1st at 9 am PT/2 pm ET. Sign up here.

👩🏻‍🏫 Advanced AI Prompting for In-House Legal - 201 on Friday, April 25th at 9 am PT/12 pm ET. Sign up here.

We’re Hiring

GC AI keeps growing across Engineering, Sales, Success, and Marketing.

Senior Full-Stack Engineer

Customer Success Manager

Founding Account Executive

Product Marketing Manager

How did you like this newsletter? What can we do better? Click to send feedback. 

❤️  I loved it!

😐 Good but could be better. 

👎 It was meh.