The AI & Legal Update #1 - 🍇 Elon's Sour Grapes

The AI, Law, and Tech Update

Read time: 5 minutes, 58 seconds. 

🧠 OpenAI, because always.

Board Ouster Investigation

Big firm Wilmer Hale investigated the events of last fall’s brief ouster of OpenAI CEO Sam Altman. The blog post about it left me unsatisfied, but did say:

• “The Board believed at the time that its actions would mitigate internal management challenges and did not anticipate that its actions would destabilize the Company.” [Wow, that prior board was not good.]

• OpenAI will add a whistleblower hotline. [Um what? Worth $40B then, and they couldn't be bothered to buy Navex until now? They must not read WGCN posts.]

New Board Members

OpenAI welcomed Sam Altman back and introduced three powerhouse women to its board. Tech, non-profit governance, and legal expertise are now stronger with Instacart CEO Fidji Simo, ex-Gates Foundation CEO Sue Desmond-Hellmann, and beltway lawyer and former Viacom president Nicole Seligman, respectively. Seligman is a lawyer who clerked for Thurgood Marshall! 

NYT Suit

OpenAI answered NYT’s suit and hired my old firm, Morrison & Foerster. The response is what you’d expect, leaning on fair use and saying that NYT “hacked” OAI by providing very specific prompts designed to elicit the AI to spit back NYT articles. Interestingly, their answer uses zero visuals. 

Elon vs OpenAI

Elon sued OpenAI for the breach of a non-existent contract. See the full complaint and my tweet thread about it.

• Tl;dr - a fun 1L contracts hypo for the world’s richest man to sue its hottest company based on him giving it’s non-profit parent money, and that non-profit later opening a for-profit subsidiary.

• OpenAI responded, calling the suit sour grapes. Their tech is good for the world and OK, it says, as shown by facts like “Albania [] using OpenAI’s tools to accelerate its EU accession by as much as 5.5 years.

Hating on marketing?

Sam Altman got some people fired up when he said “95% of what marketers use agencies, strategists, and creative professionals for today will easily, nearly instantly and at almost no cost be handled by AI.” [Not lawyers!] 

Training on YouTube

OpenAI’s CTO Mira Murati gave an otherwise good interview where he said she had no idea if Sora (the video AI) trained on YouTube or Facebook videos. People are acting surprised, but where else did they think OpenAI got video content?

🎭 Anthropic

Three new models of Claude

Anthropic introduced new models of Claude with longer context windows. I am still testing it but initial impression is it’s better for some things and weirder for others. The company’s tests are below. Full test from me soon.

📚 Regulatory

1. 🏁 Last week, Utah became the first state to pass a general AI bill: SB-149. It mandates that any interaction involving generative artificial intelligence in regulated occupations (like telemarketing) include a clear and conspicuous disclosure to the consumer that they are interacting with generative AI. Summary here.

2. 🇪🇺 The EU AI Act is almost officially law now. So many law firm breakdowns of it I don't want to favor just one. For now, see the GC AI summary here.

3. 🇮🇳 India briefly required government approval for AI models, the world freaked out, and India reversed course.

4. 🇺🇲 The Biden administration requested $3B for federal AI application development, procurement, and integration in its 2025 budget. 

🌊 Consumer & Class Actions

1. Reddit - the FTC wrote them about their deals selling data. 

2. Related - kudos to Robinhood for the excellent set of consumer disclosures about their process for getting oversubscribed Reddit IPO shares, attached. It’s clear and plain english, yet thorough. I hope I get my 19 shares. 🤞 [Narrator: sadly, she did not.]

3. A plaintiff sued GM and LexisNexis (?) after being rejected by 7 different insurers for car insurance. He alleges GM & Lexis sold the driving history from his Cadillac based on a consent to the onboard OnStar service that he never gave.

   1. See the full complaint and GC AI’s summary of his privacy and FCRA claims.

   2. Kind of makes you want to buy legal AI from LexisNexis, right? Um no. I didn't realize they were a data broker.

💼 Corporate & AI

1. 🛡️A CISO’s guide to AI: Embracing innovation while mitigating risk. A nice explainer from the CISO of tech company Zscaler.

2. ServiceNow CEO Bill McDermott said in their earnings call with investors that “there’s no price sensitivity” for customers because the AI tools are “so unbelievable.”

3. Deloitte Report on Preparing the Workforce for Ethical AI.

   1. The majority (84%) of C-level executives surveyed say their orgs are currently using AI. Nearly all (98%) surveyed companies with more than $1 billion in annual revenue are doing so.

   2. 88% say their organizations are discussing the ethical use of AI, yet just 49% report having AI guidelines in place.

   3. Editors Note: I don't love the corporate board AI charters circulated (eg asking for 5 C-suite approvals for any AI use), and want to tackle this. Email me if you want to collab.

4. See the below Economist chart showing a sharp decline in workers needed per $1B in revenue that has been making the rounds in AI circles. 

AI & Legal Practice

1. Overheard: law firm partner noting that firms can use LLMs to make money on fixed-cost work that used to be unprofitable. [Predictable.]

2. Multiple law firms have come out saying they ban AI. Good imo, helps weed out who not to hire. Also, please don’t hire law firms that give panels on AI like this counsel experienced. Boo.

🥑 Today’s random bonus

Weird one, inspired by the bad avocado toast (sacrilege!) I had while testing various AI systems for legal answers and hallucination levels. I asked 5 AIs, “what does Julia Roberts look like eating toast?” 

• Gemini didn’t know but helpfully showed a picture of her near a picture of toast, and told me she has a weakness for cookies. I knew I liked her.  

• Anthropic’s Claude apologized, then told me to try watching her movies.

• Alexa had the best answer: “Julia Roberts looks natural and radiant while eating toast, showcasing her healthy no-makeup look.” 

• Poor Siri had no idea.

• OpenAI's ChatGPT refused: "I can't create or display images of specific public figures, including Julia Roberts, due to content policy."

🚀 Product Updates

Elegant copy / paste

We heard you! You want to copy and paste GC AI answers with the pretty formatting intact, whether that be to your email client, GDocs, Word, or Slack. Now you can - just press the copy button at the bottom of any answer.

Download files you uploaded before

Download any document you’ve uploaded to GC AI. Press the new three-dot menu on any file attached to a chat message.

👩🏻‍🏫 AI Classes Updates

1. I added a March cohort of my Maven AI for Lawyers class - Thursday 03/28 at 9:30 PT. Sign up here.

2. I’ve formalized my AI corporate class offering, and soon we’ll have AI board session offering. Thanks to those who asked - from my research, literally no good board specific AI class exists. I would not hire McKinsey.

3. Follow me on LinkedIn for more announcements.

Fun times! Thank you all for your support. 

Best,
Cecilia