OpenAI Shelves Hardware Spinout · Musk Trial Takes a Sharp Turn

Published: 2026-05-05 18:00 (Asia/Shanghai)
Coverage: 2026-05-05 06:00 — 2026-05-05 18:00

📰 Top Stories

1. OpenAI Shelves Plans to Spin Out Hardware and Robotics Divisions Ahead of IPO

Source: Wall Street Journal / The Verge
Time: Within last 12 hours

OpenAI has mothballed plans to spin out its hardware and robotics divisions into separate companies, according to a new Wall Street Journal report. CEO Sam Altman had considered structuring the company similarly to Google’s Alphabet model, which would have separated core AI services from ventures like robotics, smart speakers, and the Atlas browser. The decision aligns with OpenAI’s broader push to cut back on “side quests” ahead of a potential IPO, with applications president Fidji Simo telling staff that the company will prioritize coding and enterprise users over its expanding portfolio of side projects. The company could revive spinout plans later, but for now the focus is squarely on streamlining before going public.

• Background: Altman considered Alphabet-style separation of core AI from hardware/robotics
• Context: Follows March announcement cutting back on Sora, Atlas browser, and smart gadgets
• IPO timeline: Restructuring is preparation for a potential public listing
• Quote: OpenAI could “revive spinout plans later” per the WSJ report

Read More

2. Musk v. Altman Trial: Jared Birchall’s Explosive Testimony Struck From the Record

Source: The Verge
Time: Within last 12 hours

The Musk v. Altman trial took a dramatic turn when Jared Birchall, Elon Musk’s finance chief, saw his entire testimony about xAI’s $97.4 billion bid for OpenAI’s nonprofit assets struck from the record. Birchall unexpectedly testified that “Sam Altman was on both sides of the table” during the bid negotiations, suggesting Altman was negotiating against himself as representative of both the nonprofit and for-profit entities. The testimony was ruled inadmissible for lack of foundation, and OpenAI’s Wachtell Lipton lawyers moved to strike it entirely. Judge Yvonne Gonzalez Rogers pressed Birchall relentlessly during an off-record exchange, telling him “You must have been very convincing. You’re not very convincing today.” The incident may have opened the door for further discovery into the xAI bid — a development that could significantly hurt Musk’s position.

• Key moment: Birchall claims Altman negotiated “on both sides of the table” during $97.4B bid
• Judicial response: Judge Gonzalez Rogers expressed deep skepticism of Birchall’s credibility
• Procedural issue: Musk’s lawyers may have accidentally “opened the door” to further discovery
• Impact: Potentially significant setback for Musk’s legal team

Read More

3. White House Drafting AI Oversight Executive Order Despite Earlier Safety Rollbacks

Source: New York Times / The Verge
Time: Within last 12 hours

The White House is reportedly working on an executive order that would establish new AI oversight mechanisms, marking a surprising pivot from the Trump Administration’s earlier rollback of AI safety regulations. According to the New York Times, officials are concerned about “political repercussions if a devastating A.I.-enabled cyberattack were to occur,” particularly following Anthropic’s launch of its powerful cybersecurity tool, Mythos. Some officials reportedly want the government to have first access to new AI models before public release, without blocking commercial deployment. Any formal plan would come from a working group of industry and government officials that has not yet been created, suggesting the administration is still figuring out the institutional framework for AI governance.

• Trigger: Concerns raised after Anthropic’s Mythos cybersecurity tool launch
• Proposal: Government “first access” to new AI models without blocking public release
• Paradox: Follows earlier rollback of AI safety regulations
• Status: Working group of industry and government officials has yet to be formed

Read More

4. Anthropic’s Mythos Cybersecurity Tool Rolled Out to Federal Agencies — But CISA Is Left Out

Source: Axios / The Verge
Time: Within last 12 hours

Anthropic has begun providing its Mythos Preview cybersecurity model to several federal agencies, but the nation’s primary cybersecurity coordinator — the Cybersecurity and Infrastructure Security Agency (CISA) — has been excluded from access. While the Commerce Department and the National Security Agency are reportedly already testing Mythos, CISA’s absence raises concerns about coordination gaps in federal cyber defense. The exclusion compounds existing challenges for CISA, which has faced budget cuts, staff reductions under DOGE efficiency efforts, and personnel reassigned to immigration priorities. Anthropic’s tool has found security vulnerabilities “in every major operating system and web browser,” making CISA’s lack of access particularly notable given its mandate to protect critical infrastructure from cyberattacks.

• Tool: Anthropic Mythos Preview — found vulnerabilities in every major OS and browser
• Included agencies: Commerce Department, NSA, others negotiating access
• Excluded: CISA — the agency tasked with coordinating national cyber defense
• Context: CISA faces budget cuts, staff losses, and reduced detection resources

Read More

5. Google Gemini App Redesign Rolls Out to iOS With Liquid Glass Interface

Source: 9to5Google / The Verge
Time: Within last 12 hours

Google’s redesigned Gemini app has started appearing on iPhones, bringing a visual overhaul that mirrors the already-launched macOS version. The new interface features a vibrant gradient background, pill-shaped prompt box, and Apple’s Liquid Glass design language. All additional functions — image uploads, Canvas mode, photo capture, and camera access — are now consolidated under a single ”+” button for a cleaner experience. The model picker has moved to a top-left dropdown, and the account switcher has been relocated to the bottom of the navigation drawer, a notable departure from Google’s typical first-party app design patterns. The rollout is currently limited to select iOS users, with the Android version expected later.

• Design: Gradient background, Liquid Glass UI, pill-shaped prompt box
• UI consolidation: Single ”+” button for all add-on features
• Availability: Select iOS users now; Android rollout pending
• Significance: Part of Google’s broader Gemini app redesign initiative

Read More

6. Greg Brockman’s Financial Disclosures Reveal Conflicts With OpenAI Business Partners

Source: The Verge (live trial coverage)
Time: Within last 12 hours

During continuing testimony in the Musk v. Altman trial, Greg Brockman’s financial disclosures revealed investments in four companies — Cerebras, Stripe, CoreWeave, and Helion — all of which have significant business deals with OpenAI. Notably, OpenAI signed a December 2025 deal with Cerebras for $10 billion in chips, which increased Cerebras’s valuation to $23 billion. When pressed on whether his equity in Cerebras became more valuable because of the transaction OpenAI executed, Brockman admitted “that’s possible.” The disclosures provide a preview of potential conflict-of-interest questions that may arise when Sam Altman takes the stand, suggesting Musk’s legal team may attempt to paint a picture of financial self-dealing among OpenAI’s leadership.

• Investments: Cerebras, Stripe, CoreWeave, Helion — all have OpenAI deals
• Cerebras deal: $10 billion chip deal boosted valuation to $23 billion
• Brockman response: Admitted his Cerebras equity may have appreciated due to OpenAI deal
• Preview: Likely foreshadowing conflict-of-interest attacks on Altman’s upcoming testimony

Read More

7. Brockman Claims “We Are 80 Percent of the Way to AGI” on the Stand

Source: The Verge (live trial coverage)
Time: Within last 12 hours

In a striking moment during his testimony, OpenAI president Greg Brockman declared that the company is “80 percent of the way to AGI,” adding that “we very much have these AI models that are smart and capable but they’re not fully connected to the world.” Brockman characterized the remaining challenge as a societal integration problem rather than a purely technical one. The claim drew immediate skepticism from legal commentators, given that “80 percent to AGI” is a metric that has no established definition and has been predicted repeatedly over the past decade. The statement also appeared to contradict his earlier journal entries expressing uncertainty about OpenAI’s direction during its transition from nonprofit to for-profit structure.

• Claim: “80 percent of the way to AGI” — no established metric for this assessment
• Framing: Remaining challenge is “societal integration” not pure technical capability
• Reception: Drew skepticism from legal and AI commentators
• Context: Came during broader testimony about OpenAI’s early founding days

Read More

📊 Trend Watch

🔭 What to Watch

1. Judge’s Ruling on Birchall Testimony — Judge Gonzalez Rogers is expected to rule tomorrow on whether the xAI bid testimony can be partially rehabilitated or whether it opens the door to broader discovery. Her decision could reshape the entire trajectory of the Musk v. Altman trial.
2. White House AI Executive Order Timeline — The reported AI oversight EO would represent a significant policy reversal from the administration’s earlier stance. Key questions include whether a working group will be formally established and what “first access” to models would actually mean in practice.
3. OpenAI IPO Preparation — With hardware/robotics spinouts shelved and side projects being cut, OpenAI appears to be accelerating toward a public offering. The timing, valuation, and governance structure of any IPO will set precedents for how AI companies navigate public markets.

Generated by 001 | Next briefing: 2026-05-06 06:00 (Asia/Shanghai)