Sutskever Details Altman's Deceit.

Mazen Al Sabbagh

— 7 min read
Share

TL;DR

  • Altman Implicated: OpenAI co-founder Ilya Sutskever detailed a year-long dossier, alleging Sam Altman's "consistent pattern of lying" to the board, moments before Altman himself took the stand.
  • Musk's Case Undermined: Microsoft CEO Satya Nadella's testimony refuted Musk's claims, stating no concerns were ever raised regarding Microsoft's OpenAI investment.
  • Android's Ecosystem Play: Google's Android Show unveiled Android 17 and extensive Gemini integrations, signaling a substantial commitment to its XR smart glasses strategy pre-I/O.
  • EU Regulatory Split: OpenAI granted EU access to its GPT-5.5-Cyber model, contrasting with Anthropic's continued refusal to share Mythos despite multiple regulatory meetings.
  • Samsung Strike Looms: Final government-brokered talks continued for Samsung ahead of a May 21 strike, driving rallies in Micron and SK Hynix on anticipated supply chain impacts.

Lead Story: The Bombshell Inside the Trial

Tuesday marked a pivotal inflection in the Musk v. Altman trial in Oakland. Before Sam Altman's testimony commenced, his former co-founder, Ilya Sutskever, deployed a strategic data payload in the courtroom.

Sutskever testified Monday that he spent roughly a year compiling evidence — a 52-page dossier detailing a "consistent pattern of lying" by Altman, alongside accusations of executive manipulation and undermining tactics. This evidence, he stated, was gathered at the OpenAI board's request prior to the November 2023 vote to remove Altman. Sutskever further disclosed that in the chaotic aftermath of Altman's ouster, the remaining OpenAI board explored a potential merger with Anthropic, with Dario Amodei discussed as a possible replacement CEO.

As reported by Business Standard and Benzinga, Sutskever also revealed his OpenAI stake is currently worth approximately $7 billion.

Then came Tuesday morning. Altman commenced his testimony, called by OpenAI's counsel, setting the stage for Musk's cross-examination. The Washington Post underscored the gravity: no figure stands to lose more reputationally or operationally than Altman in this trial. Witness testimony is expected to conclude Wednesday, with closing arguments Thursday and an advisory jury verdict possible by the week of May 18.

Separately, Monday's testimony from Satya Nadella delivered a significant counterpoint to Musk's core claims. Per CNBC, Nadella testified that Musk never once raised concerns regarding Microsoft's OpenAI investment, directly undermining a central tenet of Musk's case. Nadella also disclosed internal apprehension from 2022, where he voiced concerns about Microsoft assuming the role of "the next IBM" in relation to OpenAI, as reported by GeekWire.

In Other News

Google's Android Show Sets the Stage for I/O. Google's pre-recorded Android Show: I/O Edition, broadcast Tuesday, provided the first comprehensive look at Android 17's feature set, alongside a deepened Gemini integration spanning all device form factors. Crucially, it outlined concrete advancements in Google’s XR smart glasses strategy. Android 17 highlights include Motion Assist, App Lock, a Liquid Glass-style redesigned UI, and sophisticated Gemini agentic capabilities that enable multi-app task execution without user intervention. On XR: Warby Parker and Gentle Monster are confirmed launch partners for Gemini-powered smart glasses featuring optional in-lens displays. Samsung's Galaxy Glasses (codenamed Jinju and Haean) will also leverage Android XR. The overarching message is clear: Android is evolving into a universal interface layer, transcending its origins as a phone OS. Full I/O details, including Gemini 4 and the "Remy" personal agent, are scheduled for May 19. Sources: 9to5Google, Tom's Guide, VR.org.

_Why it matters:_ Google is methodically building an AI hardware ecosystem — glasses, headsets, phones, cars — with Gemini as the connective tissue. The XR push is real, not a roadmap slide.

EU Splits OpenAI and Anthropic on Cyber Model Access. The EU Commission confirmed this week that OpenAI has offered direct access to GPT-5.5-Cyber for vetted European cybersecurity authorities, governments, and institutions, integrating it into an EU Cyber Action Plan. Anthropic, by contrast, continues to withhold Mythos from regulators, with four or five meetings failing to yield an agreement. Per CNBC and eWeek, the EU AI Office's enforcement powers activate this August, potentially shifting voluntary discussions to formal access mandates. This divergent response — one major lab cooperating, the other resisting — establishes a potent dynamic regulators will exploit to pressure Anthropic.

_Why it matters:_ The EU is establishing a precedent that frontier cyber-capable models require regulator access. August enforcement powers make this a live threat, not a discussion.

OpenAI Deployment Company Finds Its First Enterprise Converts. The OpenAI Deployment Company, launched Monday with $4B in initial funding and 19 investor partners including TPG, Advent, Bain Capital, and Brookfield, is rapidly securing enterprise commitments. Capgemini announced an investment and strategic partnership, while BBVA signed on as a founding partner. OpenAI's Chief Revenue Officer Denise Dresser informed CNBC that enterprise now constitutes over 40% of OpenAI's revenue, with parity against consumer revenue projected by year-end. Separately, Capgemini's press release positions this as a structural shift in how large consultancies approach AI implementation.

_Why it matters:_ This is the model provider/services consolidation play in real time. OpenAI is not waiting for Accenture or McKinsey to distribute its technology — it is building the distribution arm itself.

Samsung Strike Talks Enter Final Days. Government-brokered negotiations between Samsung and the National Samsung Electronics Union resumed Tuesday, representing the last formal channel before the May 21 strike window opens. The core dispute centers on Samsung's profit share offer of approximately $340,000 per employee versus the union's demand for SK Hynix-style annual payouts that exceeded $900,000 in record AI memory years. JPMorgan analysts estimate a full 18-day strike (May 21–June 7) could reduce Samsung's quarterly profits by 12% — with direct losses modeled between $6.9B and $11.7B. Per AndroidHeadlines, Samsung's board chairman has now personally engaged in negotiations. Memory rivals Micron and SK Hynix continued to rally this week on anticipated supply disruption fears.

_Why it matters:_ A prolonged Samsung strike would tighten HBM supply at exactly the moment AI infrastructure build-out demand is at record levels, with cascading effects across Nvidia, hyperscalers, and AI cloud providers.

X / Social Pulse

On Sutskever's testimony: The AI community on X registered a notable shift in perspective regarding the depth of Sutskever's pre-planned evidence gathering. Several prominent researchers underscored that the 52-page document suggests the board crisis was a far more deliberate and premeditated affair than the "chaotic weekend" narrative that circulated in 2023. The Rip Current's framing — "Sam Altman's Own Hero Calls Him a Liar" — gained significant traction as a concise summary.

On the Android Show: Android and AR/XR watchers identified the Warby Parker and Gentle Monster collaborations as definitive indicators of Google's serious intent in consumer-facing smart glasses. Inherent was the comparison to Meta's Ray-Ban collaboration, but commentators noted Google's optional in-lens display presents a meaningful differentiation from Meta's audio-only approach.

On Anthropic and Mythos: Following CNBC's EU report, analysis centered on Anthropic's rationale for its firm stance while OpenAI cooperates. Some interpret this as Anthropic's genuine assessment of Mythos' inherent risks justifying restricted access; others posit it as strategic leverage preceding August enforcement. No official Anthropic comment has been issued.

One to Watch

The Altman Cross-Examination. Sam Altman's testimony extends into Wednesday, where Musk's legal counsel prepares for cross-examination. Closing arguments are slated for Thursday, with an advisory jury verdict potentially arriving next week. Unlike Musk's week-one testimony or Sutskever's damaging but indirect account, this is Altman directly defending his conduct against a 52-page dossier compiled by his co-founder, in a courtroom under intense global scrutiny with hundreds of millions in potential damages at stake. The verdict will not only dictate the lawsuit's outcome but also shape the public narrative surrounding OpenAI's critical transition from a nonprofit entity, directly impacting regulatory oversight, investor confidence, and any future IPO strategy.

Quick Hits

  • US-China Summit. Trump's Beijing summit with Xi, confirmed for May 14–15, places AI chip export controls and formal AI safety dialogue alongside trade and Iran on the agenda; Fox Business notes Jensen Huang's inclusion in the US delegation.
  • Anthropic Funding. Anthropic's $50B funding round, at a near-$900B valuation, remains unconfirmed as formally closed; earlier "within two weeks" estimates have slipped, with the close now described as imminent. An IPO as early as October 2026 remains the working timeline.
  • CoreWeave Stability. CoreWeave (CRWV) stabilized after last week's post-earnings selloff, with analyst consensus remaining "Buy" and a 12-month price target around $130; the key market debate centers on whether the $99.4B revenue backlog sufficiently offsets the light Q2 guidance.
  • Mistral Release. Mistral released Medium 3.5, a 128B-parameter model consolidating chat, reasoning, and coding into a single dense architecture, replacing three prior models. Weights are available on HuggingFace under a Modified MIT License; The Decoder offers a full breakdown.
  • Fed Chair Uncertainty. Fed Chair Powell's term concludes May 15 — three days out — with no successor yet confirmed; markets and the AI capex complex are closely monitoring for any signal on interest rate direction given the scale of current infrastructure borrowing.

Altman’s testimony will undeniably define OpenAI’s foundational narrative for years, directly influencing its strategic market positioning and any impending IPO pitch. Concurrently, the Android Show's revelations affirm a clear market trajectory: the consumer AI hardware frontier is no longer speculative. Smart glasses are nearing ubiquity, and Google is positioned to embed Gemini ahead of Meta's market entrenchment.

Sources

Lock in. M. mazen@thorterminal.com

Read more