Pivot Legal AI Briefing — May 6, 2026

Show Notes

Hosts: James Park & Priya Sharma In this episode: • Today we're covering the Musk v. Altman trial spectacle, New Mexico's massive demands from Meta, and surprising findings about AI memory and copyright... • Starting with Silicon Valley's trial of the century. James, week one of Musk versus Altman wrapped up, and I have to say, watching these proceedings o... • The transparency is unprecedented for a case of this magnitude. We saw testimony from Greg Brockman, Jared Birchall, and Musk himself. But what struck... • Right, the DeepMind CEO has become this phantom presence. From what I gathered, both sides keep referencing conversations with Hassabis from 2015 and ... • What concerns me legally is how these references to Hassabis might affect future cases. If this trial establishes that informal discussions between AI... Subscribe to the newsletter at pivotnews.ai for the full written briefing.

What is Pivot Legal — AI News Daily?

Daily AI news for legal professionals. Two hosts break down how artificial intelligence is reshaping law firms, contracts, compliance, and the justice system.

James Park: Welcome to Pivot Legal! I'm James—

Priya Sharma: —and I'm Priya. Let's get into it.

James Park: Today we're covering the Musk v. Altman trial spectacle, New Mexico's massive demands from Meta, and surprising findings about AI memory and copyright.

Priya Sharma: Starting with Silicon Valley's trial of the century. James, week one of Musk versus Altman wrapped up, and I have to say, watching these proceedings on YouTube feels like witnessing history in real time.

James Park: The transparency is unprecedented for a case of this magnitude. We saw testimony from Greg Brockman, Jared Birchall, and Musk himself. But what struck me most was how often Demis Hassabis's name came up, despite him not even being in the courtroom.

Priya Sharma: Right, the DeepMind CEO has become this phantom presence. From what I gathered, both sides keep referencing conversations with Hassabis from 2015 and 2016, particularly around AI safety governance structures. It's fascinating how someone who isn't party to the lawsuit is shaping the narrative.

James Park: What concerns me legally is how these references to Hassabis might affect future cases. If this trial establishes that informal discussions between AI leaders create binding expectations about governance, that could ripple through the entire industry.

Priya Sharma: I think that's exactly why this YouTube access matters so much. Every AI company executive is probably watching and taking notes on what not to say in casual conversations. The precedent here could fundamentally change how AI leadership communicates.

James Park: Speaking of precedent, honestly, Musk's testimony yesterday about the original OpenAI charter was surprisingly compelling. His lawyers are building a strong promissory estoppel argument.

Priya Sharma: Yeah, that tracks with what we're seeing in the documents. Though Altman's team is countering that the charter was always aspirational, not contractual.

James Park: Moving to our second story — New Mexico just escalated their fight with Meta in a way I didn't see coming. After winning that $375 million verdict, they're now seeking a $3.7 billion abatement plan.

Priya Sharma: The scale is staggering. They want Meta to fund mental health programs across the state and implement 99% CSAM detection rates. That's not just a fine anymore — it's basically asking Meta to rebuild their entire child safety infrastructure.

James Park: From a legal standpoint, this reminds me of the tobacco settlements of the 1990s. States extracted massive, ongoing payments plus fundamental changes to business practices. If New Mexico succeeds here, every state AG will be drafting similar demands.

Priya Sharma: The teen notification curbs they're demanding would essentially kill features like Instagram Stories for minors in New Mexico. I'm not sure Meta can comply with these demands without creating a completely different product for one state.

James Park: That's the leverage play though. Force Meta to choose between fragmenting their platform or negotiating a more moderate settlement. Classic litigation strategy.

Priya Sharma: Wow, that's actually wild when you think about it — one state potentially forcing a global platform to fundamentally change. The regulatory implications here are huge.

James Park: Our third story might be the most important for AI copyright law. Researchers found that Llama 3.1 70B can reproduce entire books verbatim, including all of Harry Potter and the Sorcerer's Stone.

Priya Sharma: This completely scrambles the copyright debate. For months, AI companies have been arguing their models don't actually memorize training data, just learn patterns. But here's proof that at least one major model absolutely does memorize full texts.

James Park: The legal implications are enormous. Every fair use argument relies partly on transformation — that the AI isn't just copying but creating something new. If models are storing perfect copies of copyrighted works, that defense crumbles.

Priya Sharma: I think this gives ammunition to both sides though. Publishers can point to Llama as proof of infringement, but the study also showed most other models don't exhibit this behavior. It suggests this might be a Llama-specific issue rather than an LLM-wide problem.

James Park: True, but courts tend to look at capability, not just current behavior. If LLMs can memorize books, even if most don't, that capability itself might influence how judges view the technology.

Priya Sharma: The timing couldn't be worse for Meta. They're already fighting multiple copyright suits, and now their own model is the poster child for memorization.

James Park: Honestly, I'm not buying Meta's likely defense that this is unintended behavior. When you train on internet-scale data, some memorization is inevitable.

James Park: That's your Pivot Legal briefing for May 6, 2026. I'm James—

Priya Sharma: —and I'm Priya. See you tomorrow.