Musk Becomes OpenAI Trial's Defendant.

Mazen Al Sabbagh

— 7 min read
Share

TL;DR

  • Musk v. OpenAI trial week one points to plaintiff's strategic vulnerabilities: Elon Musk's admission regarding xAI's model distillation, shifting Polymarket odds, and revelations of a covert intermediary have reshaped the court narrative, focusing more on his motivations than OpenAI's mission. (Bloomberg, WashPost)
  • Founders Fund secured its largest fund to date with a $6 billion close: The capital raise reflects a rapid deployment strategy, having expended its $4.6 billion predecessor in under a year on key AI and defense investments like Anthropic, OpenAI, Anduril, and SpaceX. (Bloomberg, TNW)
  • OpenAI activated marketing cookies by default for free ChatGPT users: This policy update formalizes advertiser data-sharing and purchase data receipt, indicating a pivot in consumer monetization strategy following reported drops in ChatGPT Plus subscriptions. (WIRED, Adweek)
  • GitHub Copilot transitions to usage-based billing from June 1: This move signals the impending obsolescence of flat-rate AI subscriptions, underscoring escalating inference costs and setting a precedent for platforms like ChatGPT, Claude, and Gemini. (GitHub Blog, PCWorld)
  • WIRED uncovered a dark-money campaign funding anti-China AI narratives via influencers: A nonprofit linked to a super PAC backed by figures from OpenAI, Palantir, and Andreessen Horowitz is orchestrating a two-phase operation to promote pro-US, then explicitly anti-China, AI sentiment on social platforms. (WIRED)

Lead Story: Musk v. OpenAI Week One: The Strategic Narrative Shifts

The inaugural week of the high-profile AI trial concluded with Elon Musk's position appearing strategically compromised. Bloomberg's assessment noted early "rough spots," a sentiment amplified by The Washington Post, which framed the case as increasingly centered on Musk's own motivations and strategic inconsistencies. MIT Technology Review's summary captured the core issue: Musk, while alleging betrayal, admitted to distilling OpenAI's models for xAI's Grok.

This admission on Day 4 proved most impactful. Under cross-examination, Musk acknowledged xAI's partial use of OpenAI's models for training, then justified it as "standard practice." This maneuver significantly eroded the moral authority of a plaintiff suing on grounds of mission deviation. Polymarket win odds for Musk subsequently dropped to 42%. While Forbes highlighted the trial's potential to redefine AI company structures, the immediate question concerns the viability of Musk's case itself.

Further revelations introduced additional complexities. WIRED reported on trial evidence suggesting Shivon Zilis, a Neuralink executive and mother to four of Musk's children, acted as an intermediary between Musk and OpenAI post-2018, querying whether she should "stay close and friendly to OpenAI to keep info flowing." Separately, the SF Business Times pointed to the unresolved question of documented conditions attached to Musk's original OpenAI donations, a critical element for the breach-of-trust claim, which wealth manager Jared Birchall's testimony failed to clarify.

CNN framed the proceedings as a contest over control of "the God machine," noting Musk's sole resonant argument concerned Microsoft's influence over AI's future, despite his broader case faltering. The Intercept offered a sharper critique, drawing a parallel between Musk's warnings of lethal AI and his companies' engagement in military AI contracts.

The trial resumes Monday, with the liability phase expected to conclude by May 21. Upcoming critical points include potential testimony from Greg Brockman and Sam Altman, and the possibility of OpenAI filing a distillation-based counterclaim.

In Other News

Founders Fund Raises Record $6 Billion After Rapid Capital Deployment. Peter Thiel's Founders Fund successfully closed its largest growth-stage fund to date, securing $6 billion. This follows the rapid deployment of a $4.6 billion predecessor fund within a year, targeting high-growth AI and defense ventures. The portfolio includes key industry players such as Anthropic, OpenAI, Anduril, and SpaceX, with average check sizes reportedly around $600 million. This capital velocity underscores the aggressive financing environment for mature private companies nearing IPO, aligning with projected public offerings for firms like Anthropic and significant valuations for Anduril. (Bloomberg, TNW)

OpenAI Initiates Default Ad Tracking for Free ChatGPT Users. OpenAI updated its U.S. privacy policy on April 30, activating marketing cookies by default for its free-tier ChatGPT users. This strategic shift, communicated to users via email, formalizes data-sharing with marketing partners and explicitly enables OpenAI to receive purchase data from advertisers. The move notably exempts paying subscribers. This pivot towards an ad-supported model comes amidst reports of an 80% decline in ChatGPT Plus subscriptions and plans for a new, lower-priced ad-supported tier. Such maneuvers, alongside efforts to establish consent-first ad mechanisms in the EU, indicate a clear reorientation of OpenAI's consumer monetization strategy toward advertising, a revealing concession for a company valued primarily for its enterprise AI dominance. (WIRED, Adweek)

The $20 AI Subscription Model Faces Erosion as GitHub Copilot Adopts Usage-Based Billing. GitHub announced a transition to usage-based billing for Copilot Pro and Pro+ starting June 1, abandoning its prior flat-rate subscription structure. The new model introduces a credit allowance, after which service halts until the next billing cycle. GitHub cited "escalating inference costs" from power users as the primary driver. This shift is interpreted by Ars Technica as an inevitable response to the disproportionate compute consumption by heavy users and, more broadly by PCWorld, as a precursor for flat-rate models across other generative AI platforms like ChatGPT, Claude, and Gemini. The economic reality is straightforward: as AI models advance and agentic workflows intensify, the fixed-price subscription becomes unsustainable against variable, high-cost compute demands. (GitHub Blog, PCWorld)

Dark Money Influence Campaign Revealed: Super PAC Funds Anti-China AI Narratives. A WIRED investigation published Friday exposed "Build American AI," a nonprofit linked to the $100 million-plus super PAC "Leading the Future," which is orchestrating a coordinated social media influence campaign. The campaign, leveraging TikTok and Instagram creators, aims to shape public opinion on AI policy, initially promoting American AI innovation and now explicitly disseminating anti-China messaging. Backers of the PAC include prominent figures from OpenAI, Palantir, Andreessen Horowitz, and Perplexity. The operation's strategic two-phase rollout, coupled with documented financial contributions from key AI industry leaders, highlights a developing pattern: internal lobbying for deregulation complemented by external influence campaigns framing competition with China as a rationale for policy direction. (WIRED, IBTimes)

X / Social Pulse

Saturday's discourse was largely framed by the Musk trial's first-week recap. Legal analysts and AI commentators converged on the perceived weakness of Musk's case, with the xAI distillation admission generating significant discussion regarding the ethics and strategic inconsistencies of a plaintiff copying the very models he accuses of misuse. The Shivon Zilis revelation sparked an additional thread of analysis, particularly concerning the dynamics of competitive intelligence within the AI sphere.

The WIRED investigation into "Build American AI" ignited a secondary wave of discussion, with critics drawing parallels between OpenAI-linked figures funding influence campaigns on AI trust, while the company simultaneously enables ad tracking for its own users. The phrase "manufactured consent for deregulation" gained traction within policy circles, pointing to a perceived strategic duality: advocate for relaxed domestic regulation while invoking external geopolitical threats and monetizing user data.

GitHub Copilot's pricing model shift catalyzed a broader debate on AI's economic sustainability. Developers noted previous encounters with credit ceilings during beta phases, solidifying the sentiment that "vibe coding isn't free." PCWorld's framing of the $20/month unlimited era's demise resonated, especially given recent tightening of usage limits across Claude Pro, ChatGPT Plus, and Gemini Advanced. Consensus points toward an inevitable industry-wide adoption of usage-based pricing, with significant implications for heavy agentic workflows.

One to Watch

Your AI Conversations Are Not Private — And Law Enforcement Exploits This Gap.

CNN's Saturday investigation revealed a critical vulnerability in the perceived privacy of AI interactions: ChatGPT conversations are becoming a routine evidentiary source in criminal investigations. The report details multiple instances where subpoenaed AI chat histories were introduced in court, ranging from murder cases involving alleged research into methods to establishing intent in other jurisdictions. The underlying legal framework is notably insufficient; no specific statutory protections for AI conversations currently exist in U.S. law. Existing privacy statutes, designed for traditional digital communications, fail to adequately cover the interactive nature and depth of AI sessions.

The timing of this report is significant, arriving shortly after OpenAI enabled marketing cookies by default and formalized advertiser data-sharing. While OpenAI maintains that "your conversations with ChatGPT are private," this definition appears to be narrowly scoped, excluding law enforcement access via subpoena. As users increasingly engage chatbots for sensitive, personal, and professional advice, the chasm between user privacy expectations and legal reality widens. The absence of transparency reports from major AI companies regarding law enforcement data requests is becoming increasingly indefensible in an environment demanding greater accountability and trust architecture.

Quick Hits

  • Flourish, a brain-inspired AI efficiency startup, seeks $500 million at a $2.5 billion valuation. Founded by Internet Explorer creator Thomas Reardon, the venture aims to mitigate AI's burgeoning power consumption through neuroscience-inspired architectures, with Lux Capital and GV leading the round. (Bloomberg)
  • Oracle was swiftly added to the Pentagon's classified AI deal list, expanding the roster to eight firms with IL6/IL7 network access, notably excluding frontier lab Anthropic from this elite group. (Breaking Defense)
  • A Povaddo survey indicates 90% of U.S. and European policy insiders believe AI regulation is essential and governments are currently falling short, with four in ten U.S. respondents identifying AI as an existential threat. (PR Newswire)
  • Fed Vice Chair Bowman highlighted Anthropic's Mythos as demonstrative of AI's dynamic nature, signaling the need for adaptable regulatory oversight as the Fed crafts a new playbook for bank AI integration. (Bloomberg)
  • Despite a surge in AI-generated music on streaming platforms, listener acceptance is declining, a new study reveals growing discomfort amidst increasing volume. (NPR/Alabama Public Radio)

The week concluded with Big Tech earnings reinforcing the tangible reality of AI revenue, even if a significant portion of Google's and Amazon's reported "AI profits" stemmed from revaluing their Anthropic stakes rather than core operational gains. The Pentagon solidified its AI supplier base, pointedly excluding Anthropic, while the Musk trial unveiled strategic contradictions from a plaintiff whose own ventures mirrored the alleged mission drift. Markets reacted with record highs. Concurrently, the vanguard of AI development is simultaneously enabling user tracking for advertisers, deploying dark-money influence campaigns to mitigate regulation, and framing geopolitical competition as the primary external threat. The week ahead brings the second phase of the trial—with key testimonies anticipated—the allocation window for the Anthropic funding round, and trading after April's record rally. The essential inquiry is not whether AI constitutes a bubble, a notion already dismissed. Rather, it concerns the economic sustainability of serving hundreds of millions of users at fixed low prices and whether the industry's public trust can endure the dissonance between its aspirational rhetoric and its tactical commercial behavior. GitHub's latest move offers a direct answer to the former. WIRED's investigation suggests the latter is already facing substantial erosion.

Sources

Lock in. M. mazen@thorterminal.com

Read more