The Coming AI Rules Battle

March 23, 2026 · Episode Links & Takeaways

Author

AI Daily Brief
March 23, 2026

HEADLINES

OpenAI Is Doubling Headcount

OpenAI plans to double headcount to around 8,000 this year, requiring roughly a dozen hires per day — a sharp reversal from Sam Altman's January statement that the company planned to "dramatically slow down" hiring. The new employees will span product, engineering, research, and sales, with a notable new function: "technical ambassadors" who help enterprises actually implement their tools. An unnamed OpenAI executive told the FT that AI coding tools had "opened up entirely new lanes of things we can do. All of a sudden, the company kind of rotated on its axis." OpenAI's own Adam GPT summed it up: "It feels like we are top of the third inning. The models aren't the problem. They're smart enough now. Now it's about applying them at scale."

FedEx Is Training All 400,000 Employees on AI

FedEx has begun delivering bespoke AI training to its entire workforce, from drivers to executives. The initiative began in December, is delivered in partnership with Accenture, and is tailored to individual roles — including what FedEx calls "communities of practice" with hackathons and use case sharing. The full C-suite recently took two days off to visit Silicon Valley and meet with vendors, which CDIO Vishal Talwar described as a level of institutional humility he'd never seen before. What's notable here isn't the PR — it's that the training sounds genuinely bespoke and comprehensive at a time when most organizations haven't even started.

HSBC Weighs Cutting 20,000 Jobs as AI Takes Over Back Office

Bloomberg reports HSBC is mulling a 10% headcount reduction — up to 20,000 jobs — as the bank bets on AI to cut middle and back office functions, spread over three to five years. Bloomberg Intelligence predicted last year that global banks will eliminate 200,000 positions over a similar horizon, and a Business Insider survey of banking CTOs put average expected workforce reductions at 3%. The contrast with FedEx couldn't be sharper: one company is investing in upskilling 400,000 people, while another is preparing to eliminate them.

Mark Zuckerberg Is Building an AI Agent to Help Him Be CEO

The Wall Street Journal reports that Zuckerberg is building a personal AI agent focused on making information sharing more efficient throughout Meta — surfacing insights that would otherwise require traversing layers of management. The agent reflects a broader transformation: Meta is stripping management layers, installing flatter teams, and simultaneously rolling out agents to turbocharge the effort. Meta has two agents deployed org-wide: "My Claw" (an OpenClaw variant with access to chat logs and work files) and "Second Brain" (built on Claude, functioning as an agentic knowledge base and "AI chief of staff" for every employee). Agents are already talking to each other to resolve issues without involving their human owners. AI use is now graded in performance reviews, layoff anxiety is real, but others describe the atmosphere as reminiscent of the early "move fast and break things" era.

MAIN STORY

The Coming AI Rules Battle

AI was always going to become a major political issue in 2026 — midterms are coming, AI is touching more people's lives every day, and the economic anxiety around it is real and growing. Blue Rose Research's David Shore put it plainly: AI is rising in importance faster than any other issue they track, although it's only ranked 29th out of 39 issues right now. The White House's new national AI legislative framework is the opening move in what will be a long, contentious, multi-dimensional public negotiation — and understanding what's in it, who supports it, and who opposes it is essential context for everything that follows.

The Political Backdrop
"Not a great starting place for major disruption to the labor market."
Blue Rose Research finds AI is rising in political importance faster than any other tracked issue — already ranking above climate change, abortion, and guns in terms of rate of ascent, even while still 29th out of 39 overall. The economic context matters enormously: 61% of Americans say life has gotten less affordable in the past year. More than 50% are concerned they or a family member will lose a job in the next year, and 56% specifically attribute that fear to AI. 77% are concerned about entire industries being eliminated. When asked to choose between "fund new jobs even if it limits AI company profits" vs. "keep innovating for American dominance," funding jobs won among Trump voters two to one. People are not interested in UBI as an answer — they want good paying jobs, by a three to one margin across every demographic.

David Shor (X) Blue Rose Research Polling Data

The White House Framework
"The polar opposite of Blackburn's 291 pages."
The framework is a four-page document with six named points: protecting children and empowering parents; safeguarding American communities (including the Ratepayer Protection Pledge requiring AI companies to fund their own energy infrastructure); respecting IP rights while letting courts determine training data legality; preventing censorship and protecting free speech; enabling innovation through existing sector-specific regulators rather than a new central AI body; and educating Americans. A seventh section — the longest and most detailed — addresses preempting state laws. The ratepayer section is arguably the least controversial, with virtually every major AI company having already signed on. The most delicate section is IP: the White House affirms that training on copyrighted material doesn't violate copyright, while proposing voluntary licensing frameworks for rights holders to negotiate collectively without antitrust exposure.

The State Laws Context
"States are moving because the federal government hasn't."
California signed SB 53 in September — the first US frontier AI law — covering whistleblower protections, incident reporting, and chatbot guardrails for minors. New York signed the RAISE Act in December with similar reporting requirements, million-dollar penalties, and a new oversight office in the Department of Financial Services. Both OpenAI and Google have publicly said that absent a federal framework, states like California and New York represent "manageable" or "emerging" models to align around. Meanwhile, political candidates are making AI regulation a campaign platform: New York's Alex Bores, who sponsored AI safety legislation, is now being targeted by AI industry super PACs as he runs for federal Congress. A separate proposed New York bill would bar chatbots from providing legal or medical advice and allow users to sue for bad AI guidance — something the White House's free speech section would effectively prohibit.

The Responses
"An excellent foundation" — or "Mad Max for the AI industry."
The White House's biggest political challenge may not be from the left. Bannon's War Room account amplified a characterization of the framework as enabling a "transhuman" and "profoundly anti-human" agenda — suggesting the populist right is not fully on board. Senator Blackburn, whose 291-page draft claimed to represent the White House position, said she "welcomes" the White House to the discussion but still believes her own bill is "the solution America needs." Senator Cruz appears to be positioning himself as the White House's congressional champion against Blackburn. On the Democratic side, Representative Gottheimer's statement — the framework "still has a long way to go" — actually leaves significant room for collaboration, which is more than a flat rejection. Former Trump AI advisor Dean Ball's take is the right frame: this is not a policy document, it's the opening move in a long multi-dimensional negotiation. "You must read it that way."