[AI WEEKLY NEWS RUNDOWN] The Labor Squeeze, Apple’s Intel Deal, and OpenAI’s Smartphone (May 03 – 10 2026)

🎧 Listen Ads-Free: Subscribe to DjamgaMind at https://podcasts.apple.com/us/channel/djamgamind/id6760446113

#DJAMGAMIND

Summary: In this weekly briefing, we analyze the “Automation Squeeze.” We deconstruct the massive wave of white-collar layoffs as Cloudflare and Coinbase cut up to 20% of their staff to implement “agentic AI-first” operations. We explore the fracturing hardware supply chain, highlighted by Apple’s preliminary chip deal with Intel and a collapse in global motherboard sales due to soaring RAM prices. We also dive into the courtroom drama of Mira Murati testifying against Sam Altman, the surprising compute truce between Elon Musk’s SpaceX and Anthropic, OpenAI’s push to build 30 million AI smartphones by 2027, and a landmark Harvard study proving AI can outperform ER doctors in diagnostic triage.

Important Topics:

• The Agentic Labor Purge: Cloudflare cuts 1,100 jobs (20%) and Coinbase cuts 700 jobs (14%), explicitly citing the shift to AI-driven productivity and flattened org charts.

• Apple & Intel Chip Deal: Apple signs a preliminary agreement with Intel for US-based chip manufacturing to hedge against TSMC capacity limits.

• Motherboard Market Collapse: Asus, Asrock, and MSI slash 2026 projections by 30% as the AI data center buildout causes a severe DDR5 memory shortage (”RAMageddon”).

• Murati Testifies Against Altman: Former OpenAI CTO Mira Murati testifies in federal court that Sam Altman lied about safety clearances and pitted executives against each other.

• OpenAI Fast-Tracks Smartphone: OpenAI partners with MediaTek to begin mass-producing an “AI agent phone” in early 2027, targeting 30 million units.

• Musk & Anthropic Truce: SpaceX(AI) signs a massive deal to lease the Colossus 1 supercluster to Anthropic, doubling Claude’s usage limits.

• AI Beats ER Doctors: A Harvard study shows OpenAI’s o1 model diagnosing 76 real ER cases more accurately than two attending physicians using unstructured notes.

• Enterprise Finance Push: Anthropic ($1.5B joint venture) and OpenAI ($4B deployment company) launch massive consulting arms to force AI into Wall Street and mid-sized businesses.

• DeepMind UK Unionizes: Google DeepMind workers in London vote to unionize to block the lab from supplying military AI to the US and Israel.

• Amazon Attacks UPS: Amazon launches “Supply Chain Services,” opening its 100-plane, 80,000-trailer network to B2B shipping.

🔗 RESOURCES

The AI landscape moves faster than a hallucinating LLM on a double espresso, which is why I’ve done the heavy lifting for you. Stop scrolling through generic “Top 10” lists and head over to the AI Executive Toolkit at https://djamgamind.com/toolkit

• Find REMOTE AI Jobs (Mercor): Apply Here

• Google Workspace: Professionalize your firm’s infrastructure with secure, cloud-based collaboration and branded communication: https://referworkspace.app.goo.gl/Q371

• AI Learning App Recommendation: https://apps.apple.com/ca/app/ai-ml-tutor-pro/id1610947211

⚗️ PRODUCTION NOTE: We Practice What We Preach.

AI Unraveled is produced using a hybrid “Human-in-the-Loop” workflow.

Apple and Intel reach preliminary chip deal

• Apple has signed a preliminary agreement with Intel to manufacture some of the chips that go into Apple devices, according to The Wall Street Journal, ending more than a year of intensive talks between the two companies.

• It’s not yet clear which Apple products Intel will make chips for, but earlier reports from Ming-Chi Kuo and Jeff Pu suggested M-class chips for Macs and iPads in 2027, with iPhone chips possibly following in 2028.

• The US government played a key role in the deal, with President Trump personally advocating for Intel to Tim Cook at the White House, which fits the administration’s push to secure new business for the partly US-owned chipmaker.

Google adds more website links to AI Overviews

• Google is rolling out changes to AI Overviews and AI Mode that add more links to outside websites, including a new “Further Exploration” section at the bottom of answers that suggests related articles in a bullet point list.

• The “Further Exploration” box may also surface follow-up bait questions, and in the example Google shared, a search about urban green spaces points users to content covering specific projects in New York and Singapore.

• A separate “Expert Advice” section will pull snippets from news stories, reviews, public forums, and social media posts tied to the query, with each entry linking out so readers can jump to the full conversation.

Sales of motherboards are collapsing

• Skyrocketing RAM prices tied to AI data center demand are dragging down motherboard sales, with Taiwan’s four biggest makers — Asus, Asrock, Gigabyte, and MSI — sharply cutting their 2026 shipment projections by roughly 25 to 30 percent.

• Asus expects to ship only 10 million motherboards in 2026, down from 15 million last year, while Asrock projects a drop from 4.4 to 2.7 million as DIY builders skip purchases they can’t fully equip.

• The DDR5 shortage has rippled outward, pushing Valve to likely delay its Steam Machine, forcing Apple to raise Mac prices, hiking console costs, and reportedly leading Nvidia to cancel its RTX 50 Super lineup.

ByteDance Bids for Video Leadership

As OpenAI prepares to shut down Sora, ByteDance made its own video generation model available to hundreds of millions of users.

What’s new: ByteDance added Seedance 2.0, its multimodal video generator, to its popular video-editing app CapCut. Launched earlier this year in China, the model now reaches paying CapCut users in Southeast Asia, Latin America, Africa, the Middle East, parts of Europe, Japan, and the United States.

• Input/output: Text, images, audio, and video in (up to 3 video clips, 9 images, and 3 audio clips), synchronized video and audio out (4 to 15 seconds at 480 or 720 pixels on the shorter edge in 6 aspect ratios: 21:9, 16:9, 4:3, 1:1, 3:4, and 9:16)

• Features: Lip-synced dialogue in multiple languages, ambient sound, music, multiple camera shots with cuts in a single clip, camera and lighting controlled by prompts, outputs marked by invisible watermark, blocking of input images that contain real faces or copyrighted characters (via CapCut)

• Performance: Within top two on Arena AI and Artificial Analysis video leaderboards

• Availability/price: Via CapCut (Jianying in China) paid tier, Dreamina web interface, API via the ByteDance services BytePlus and Volcengine, and third-party providers including Higgsfield.ai for $0.30 per second of output (720 pixels, audio included) or $0.24 per second for faster processing by SeeDance 2.0 Fast

• Undisclosed: Architecture, parameter count, training data and methods

How it works: Seedance 2.0 extends ByteDance’s earlier work from synchronous generation of audio-video streams in parallel to joint generation within a unified system. ByteDance’s launch announcement characterizes the architecture as “sparse.”

How Nvidia Uses AI to Design Chips

Nvidia’s chief scientist dreams of telling an AI model to design a new GPU, then skiing for a couple days while the system does the job. He outlined Nvidia’s progress toward that goal and how far it has to go.

What’s new: Bill Dally, who leads roughly 300 researchers at Nvidia, described AI’s growing role in designing the company’s chips in a conversation with his Google counterpart, Jeff Dean, onstage at Nvidia’s GTC conference in mid-March. His examples (starting in the video at around 24 minutes) ranged from a reinforcement learning system that lays out a chip’s building blocks to large language models trained on decades of proprietary documents.

How it works: Nvidia applies AI at five stages of chip design: laying out components, designing arithmetic circuits (components that perform math on binary numbers, like adders and counters), general engineering assistance, verifying finished designs, and exploring novel layouts.

AI at Work, Quantified

Half of workers in the United States used AI at work at least a few times last year, a sign of steadily rising AI adoption in U.S. workplaces.

What’s new: Most U.S. workers who used AI found that it boosted their productivity, according to a poll conducted by Gallup, an organization that surveys public opinion on a wide variety of topics. Respondents were most likely to use the technology when it fit into the way they worked and their employers supported it. Still, a sizable portion of employees and employers are holding out.

How it works: Gallup surveyed 23,700 U.S. employees between February 4 and February 19 on a range of questions related to AI and work. They explored the technology’s impact on productivity, whether it is changing workflows, and whether organizations are supporting and integrating it. Some employees remain skeptical of AI, but the findings suggest that AI improves productivity and plays a larger role in organizations that support its use and provide suitable tools.

• Regular AI use is rising steadily. For example, 13 percent of respondents said they used AI daily, and 28 percent used it a few times a week. These figures are up from 4 percent and 11 percent respectively in 2023. At the organization level, two in five workers said their employers had introduced AI tools into the workplace and a quarter of companies had clear AI strategies.

• AI is boosting productivity but doesn’t yet substitute for established processes. Within companies where AI was used, 65 percent of employees said it improved their productivity, while 31 percent said it had changed the ways they worked. Only 7 percent of respondents who worked in organizations where AI was used disagreed that AI had affected how they worked.

• Managerial support influences employees’ behavior and outlook. Employees with strongly supportive managers in organizations that used AI were more likely to use AI and agree it had transformed their work.

• Low users and non-users widely indicated a desire to keep doing the work they do now. Other common barriers to AI adoption included ethical concerns, data privacy, and a belief or experience that AI wasn’t useful.

Behind the news: According to some accounts, AI’s impact has been disappointing relative to the promises made by tech evangelists. “AI is everywhere except in the incoming macroeconomic data,” such as metrics that gauge employment, productivity, and inflation, writes Torsten Slok, chief economist at the investment firm Apollo. By other accounts, evidence is mounting that AI is impacting the job market. Research published by Stanford economists last year found that employment was declining for workers whose jobs may be affected by AI, such as software developers and customer-service representatives.

Why it matters: The Gallup results suggest that workers use AI to help them do their jobs, not to do their jobs for them. This can be good both for workers, who may be freed of monotonous tasks, and their employers, which may gain productivity. But AI has the potential to automate some positions entirely. The jury is still out regarding whether AI-driven productivity gains will reduce or increase overall employment.

Cloudflare cuts 1,100 jobs citing AI

• Cloudflare is laying off more than 1,100 workers, about 20% of its staff, saying that agentic artificial intelligence has “fundamentally changed” how the company operates and reshaped which roles it needs going forward.

• The cloud company beat analysts’ expectations in its first-quarter earnings reported Thursday, but shares still dropped 18% in extended trading after the workforce reduction was disclosed alongside the results.

• CEO Matthew Prince called the cuts the “right decision” on the earnings call, and Cloudflare said its own use of AI has jumped more than 600% over the past three months under an “agentic AI-first operating model.”

OpenAI launches 3 new voice models

• OpenAI has released three new voice models — GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper — designed to reason, translate across languages, and transcribe speech in real time through the company’s Realtime API and Playground.

• GPT-Realtime-2 brings reasoning on par with GPT-5, expands the context window from 32,000 to 128,000 tokens, calls multiple tools in parallel, and uses preambles like “one moment” instead of going silent during delays.

• GPT-Realtime-Translate covers more than 70 input languages and 13 output languages at $0.034 per minute, while GPT-Realtime-Whisper handles streaming transcription for live captions at $0.017 per minute, with Deutsche Telekom already testing voice-to-voice support.

Google unveils Whoop-like screenless Fitbit Air

• Google has launched the Fitbit Air, a $100 screenless wearable in the style of Whoop that tracks heart rate, AFib, blood oxygen, sleep stages, and heart rate variability around the clock.

• The band weighs 12 grams, runs for up to a week on a charge with five-minute fast charging for a full day, is water-resistant to 50 meters, and pairs with the Pixel Watch.

• Google also rebranded the Fitbit App as the Google Health app and opened up Google Health Coach, its Gemini-powered trainer and sleep advisor, to Google Health Premium subscribers, with the Fitbit Air shipping May 26.

Anthropic strikes massive deal with SpaceX

• SpaceX, which recently took over xAI, has signed a deal giving Anthropic access to computing power at the Colossus data center, despite Elon Musk previously calling the Claude maker “Misanthropic,” “evil,” and an enemy of Western Civilization.

• Anthropic said the agreement lets it double Claude Code’s five-hour rate limits for Pro, Max, Team, and Enterprise plans, drop the peak-hours cutoff for Pro and Max subscribers, and raise API rate limits for Claude Opus.

• Musk also announced that xAI will be dissolved and folded into “SpaceXAI,” while Colossus 1 opens to Anthropic and SpaceX shifts focus to Colossus 2, which aims for 1 gigawatt of power but isn’t running at full capacity yet.

Murati tells court she couldn’t trust Sam Altman

• Former OpenAI CTO Mira Murati told a court in a video deposition during the Musk v. Altman trial that she did not trust CEO Sam Altman, saying he lied to her about safety standards for a new AI model.

• Murati testified that Altman falsely claimed OpenAI’s legal department had cleared a GPT model from going through the deployment safety board, but general counsel Jason Kwon contradicted that, so she sent the model through the board anyway.

• Murati agreed Altman pitted executives against each other and undermined her, echoing cofounder Ilya Sutskever’s 52-page memo, though she still criticized the board’s firing of Altman, saying OpenAI risked falling apart before she left in 2024.

Apple may drop $599 MacBook Neo

• Apple is weighing whether to cut the $599 MacBook Neo with 256GB storage, a move that would push the laptop’s starting price up by $100 without raising the cost of any single configuration, according to Tim Culpan.

• The pricing squeeze follows Apple telling suppliers to double production to 10 million units after stronger-than-expected demand, which drained A18 Pro chip inventory and ran into limited 3nm capacity at TSMC due to AI orders.

• The first Neo batch used lower-bin A18 Pro chips with one GPU core disabled, but a fresh run would yield more fully working chips at a higher per-unit cost, and Apple may instead add new colors to soften a price hike.

Google shuts down Project Mariner AI agent

• Google has discontinued Project Mariner, the experimental AI agent built to carry out tasks across the web, with its landing page confirming the May 4th, 2026 shutdown and noting the technology moved into other Google products.

• Launched in December 2024 and later updated to handle up to 10 tasks at once, Project Mariner’s features were folded into Gemini Agent, which can archive emails or book hotels, and into AI Mode for search.

• Google recently showed a Chrome feature called “auto-browse” that handles multi-step tasks like researching flight costs, and the Mariner shutdown may clear space for new AI tools debuting at I/O starting May 19th.

Spotify opens personal podcasts to AI agents

• Spotify is now letting people bring AI-generated personal podcasts into its app through a new beta CLI tool that works with coding agents like OpenAI’s Codex, Anthropic’s Claude Code, and OpenClaw.

• The imported podcasts, which can cover things like class note summaries or calendar briefings, show up in the user’s Spotify library for private listening and cannot be accessed by other Spotify users.

• To set it up, users visit the tool’s GitHub page, log in to Spotify through a browser, then prompt their agent to generate a podcast on a topic and save it, receiving a Spotify link back.

Apple pays $250M over Siri AI delays

• Apple has agreed to pay $250 million to settle a class action lawsuit over its delayed rollout of the “more personalized Siri” features that were first shown at WWDC 2024, without admitting any wrongdoing.

• The settlement works out to about $25 per eligible device bought in the US between June 10, 2024 and March 29, 2025, though payouts could climb as high as $95 depending on claim numbers.

• The lawsuit accused Apple of promoting AI features “that did not exist at the time, do not exist now, and will not exist for two or more years,” saturating airwaves to build consumer expectations around the iPhone’s release.

OpenAI plans 30 million phones in two years

• OpenAI is fast-tracking an “AI agent phone” and could begin mass production in early 2027, with analyst Ming-Chi Kuo predicting 30 million units will be built between 2027 and 2028 if plans hold.

• The device will run on a customized version of MediaTek’s Dimensity 9600 chipset, with Kuo saying MediaTek will likely be the sole processor supplier, while Qualcomm and Luxshare are also working with OpenAI.

• Kuo said the phone will use a dual-NPU architecture for heterogeneous AI compute, splitting tasks like image enhancement and object detection between two AI processors, plus an image signal processor with enhanced high dynamic range.

Apple opens iOS 27 to rival AI models

• Apple will let iPhone, iPad, and Mac owners pick from several outside AI models to run features built into iOS 27, iPadOS 27, and macOS 27, with the shift planned for this fall, according to people familiar with the matter.

• The third-party AI services will handle tasks like generating and editing text and images across Apple’s software, expanding the company’s strategy to turn its devices into a broad AI platform rather than relying on a single provider.

• The sources who described the plans asked not to be identified because the details are private, and the article does not name which outside AI models Apple intends to offer users when the new operating systems ship.

Publishers sue Meta over AI book piracy

• Five book publishers — Hachette, Macmillan, McGraw Hill, Elsevier, and Cengage — along with author Scott Turow, have filed a class action lawsuit accusing Meta and Mark Zuckerberg of pirating copyrighted books to train the Llama AI platform.

• The complaint claims Zuckerberg personally authorized and encouraged the infringement, saying Meta reproduced and distributed millions of copyrighted works without permission or compensation, knowing the conduct violated copyright law.

• Meta spokesperson Dave Arnold pointed to past rulings that training AI on copyrighted material can qualify as fair use, echoing a recent Anthropic case where a judge rejected copyright infringement but floated piracy as a separate path to damages.

Google tests Remy 24/7 AI agent

• Google is testing an AI agent codenamed Remy that aims to turn the Gemini app into a 24/7 personal assistant for work, school, and daily life by taking actions on a user’s behalf.

• According to Business Insider, Remy is built deep into Google’s ecosystem and can monitor things that matter to users, handle complex tasks proactively, and learn preferences over time, positioning it against OpenClaw, which OpenAI acquired.

• Remy is currently in a “dogfooding” stage with employees testing it internally, and while no launch date is set, Google’s I/O event on May 19-20 could serve as the debut for the new agent.

OpenAI launches GPT-5.5 Instant for ChatGPT

• OpenAI has rolled out GPT-5.5 Instant as the new default model in ChatGPT, replacing GPT-5.3 Instant, with the company saying it cuts down hallucinations in sensitive fields like law, medicine, and finance while keeping low latency.

• The model scored 81.2 on the AIME 2025 math test versus 65.4 for the older version, and hit 76 on the MMMU-Pro multimodal reasoning benchmark compared to the predecessor’s 69.2.

• GPT-5.5 Instant can pull from past conversations, files, and Gmail for personalized answers, starting on web for Plus and Pro users, and ChatGPT will now show memory sources that users can delete or correct across all models.

Apple explores Intel and Samsung for US chips

• Apple is in early talks with Samsung and Intel to make chips in the US, hoping to secure backup options if TSMC cannot keep up with its demand for main chipsets in its devices.

• TSMC handles roughly three-fourths of global chip fabrication, but the surge in AI chip orders is straining its capacity, pushing phone makers like Qualcomm to already lean on Samsung for parts of the Snapdragon 8 Elite Gen 6.

• Apple executives have reportedly toured Samsung’s $17 billion fabrication plant in Taylor, Texas, which is expected to come online by late 2026, though Bloomberg notes the discussions with both chipmakers are far from finalized.

White House considers tighter regulation of new AI models

• The Trump administration is weighing an executive order that would set up a working group of tech executives and officials to review new AI models before release, a sharp turn from its earlier hands-off stance on the technology.

• The shift followed Anthropic’s announcement of Mythos, a model so strong at finding software security flaws that the company withheld it from the public, prompting White House worries about a possible AI-enabled cyberattack on its watch.

• Susie Wiles and Scott Bessent have taken over AI policy after David Sacks left in March, and they are also trying to repair ties with Anthropic after the Pentagon cut off its technology in a fight over a $200 million contract.

Coinbase cuts 14% of staff citing AI

• Coinbase is laying off roughly 700 workers, about 14% of its staff, as CEO Brian Armstrong pushes a restructuring tied to crypto market volatility and a bigger push to use AI tools across the company.

• The reorg flattens the company to five layers below the CEO and COO, lets leaders manage more than 15 direct reports, and asks managers to contribute more work themselves rather than just oversee others.

• Coinbase will try “one-person teams” blending engineering, design, and product management roles using AI tools, and expects to take $50 million to $60 million in severance costs, according to an SEC filing.

Brockman discloses $30 billion OpenAI stake

• OpenAI president Greg Brockman testified Monday that his equity stake in the company is worth close to $30 billion, while also revealing financial links to Sam Altman through investments in Altman-backed ventures and his family office.

• Court records showed Altman gave Brockman an interest in his personal investment fund in 2017, which Musk’s lawyers argue may have hurt Brockman’s independence, citing an email from Musk associate Jared Birchall.

• Brockman also acknowledged holdings in AI chipmaker Cerebras and fusion startup Helion Energy, both tied to Altman’s investment network, as Musk’s lawsuit seeks $150 billion in damages and OpenAI’s return to nonprofit status.

DeepMind UK staff unionize over military AI deals

• Google DeepMind workers in London have voted to unionize, asking the company to recognize the Communication Workers Union and Unite the Union as joint representatives in a push to stop the AI lab from supplying its tech to the US and Israeli militaries.

• The effort started in February 2025 after Alphabet dropped a pledge against using AI for weapons and surveillance, and gained urgency after a reported Google deal letting the Pentagon use its AI for “any lawful government purpose.”

• If unionization succeeds, staff plan to demand Google exit its contract with the Israeli military, share more details on how its AI products will be used, and offer some assurance about layoffs driven by automation.

Meta deploys AI to detect underage users

• Meta is rolling out AI age-detection technology to Facebook users in the U.S. for the first time, along with 27 European Union countries and Brazil, as part of new measures announced during a child safety trial.

• Parents in the U.S. on Facebook and Instagram will get a notification linking to a blog post explaining how to check and confirm their teens’ ages, sent to all users Meta has identified as a parent.

• The announcement lands during the second phase of a New Mexico trial where the state seeks $3.75 billion in damages and policy changes, prompting Meta to threaten pulling Facebook, Instagram, and WhatsApp from the state.

Ask.com shuts down after 25 years

• Ask.com has officially closed its doors after 25 years online, with parent company IAC confirming on May 1, 2026 that it shut down its entire search business, ending one of the web’s earliest recognizable search brands.

• Born in the late 1990s alongside Google Search, the platform started with a question-and-answer format and a butler mascot named Jeeves before rebranding to Ask.com, but it steadily lost ground to Google’s ranking systems.

• In a farewell message, IAC said “a very great search must come to an end” and thanked its engineers, designers, and the millions of users, noting that Ask’s natural-language approach foreshadowed today’s conversational search and AI tools.

AI beats doctors on diagnoses in Harvard study

• OpenAI’s o1 model matched or beat board-certified emergency room physicians on diagnosis, triage and next-step care decisions in a new Science study, based on six experiments using real data from a Massachusetts medical centre.

• The model stood out during early-stage triage, where it handled uncertainty better than doctors by making stronger use of unstructured notes and partial information, though both humans and AI improved as more data came in.

• Researchers, including Harvard’s Arjun Manrai and commentators from Flinders University, warned that AI cannot read visual, body language or auditory cues, and called for prospective clinical trials covering safety, equity and cost-effectiveness before wider use.

Anthropic nears $1.5B Wall Street venture

• Anthropic is close to sealing a roughly $1.5bn joint venture with Blackstone, Hellman & Friedman, Goldman Sachs, and General Atlantic that will push Claude into the portfolio companies owned by those Wall Street firms.

• Anthropic, Blackstone, and Hellman & Friedman are each anchoring the deal at about $300m, Goldman Sachs joins as a founding investor at roughly $150m, and General Atlantic and others cover the rest.

• The venture follows OpenAI’s DeployCo, which drew $4bn from five PE firms last month with a 17.5 per cent annualised return guarantee, while Anthropic’s smaller structure has no publicly reported guaranteed returns.

Amazon declares war on UPS and FedEx

• Amazon is opening its freight network to businesses in retail, healthcare, manufacturing and other industries through a new service called Amazon Supply Chain Services, letting them move, store and deliver goods by ocean, road, rail and air.

• The offering taps Amazon’s fleet of more than 100 cargo planes, over 80,000 trailers and 24,000 intermodal containers, plus warehouses, and includes distribution, fulfillment, parcel shipping, two-to-five-day delivery, warehousing and inventory forecasting.

• The push targets the business-to-business shipping market, a high-margin segment for logistics firms, and mirrors the playbook of Amazon Web Services, which started in 2006 as internal infrastructure before becoming the biggest cloud provider.

Musk texts Brockman seeking OpenAI settlement

• OpenAI’s lawyers have asked a federal judge to let Greg Brockman testify about an April 25 text in which Elon Musk floated a settlement, then warned Brockman and Sam Altman would become “the most hated men in America.”

• The defense says the text shows motive and bias rather than settlement value, citing Federal Rule of Evidence 408 and arguing Musk is using the Oakland lawsuit to attack a competitor after launching xAI.

• Judge Yvonne Gonzalez Rogers, not the nine-person advisory jury, will decide liability and remedies, which could include removing Altman and Brockman from leadership, disgorgement to the charity, and unwinding OpenAI’s for-profit conversion.