AI Companies Are Hiring More

Jul 2, 2026 · Episode Links & Takeaways

Author

AI Daily Brief
July 02, 2026

HEADLINES

OpenAI Proposes Handing the US Government a 5% Stake

Those conversations about AI companies getting more tied up with the government are getting a lot more real. OpenAI has floated giving 5% of the company to a sovereign wealth fund structured like the Alaska Permanent Fund, a stake worth roughly $42 billion at current valuations, with it still unclear whether the government would buy in or receive it as a gift. OpenAI has also proposed that other leading labs, including Anthropic, Google, and Meta, contribute 5% of their own equity, though none have signed on. Discussions are still "conceptual" and might require an act of Congress, but the Overton window on these deals is clearly shifting, and Micron's $250 million donation to Trump accounts the same day is another data point in this strange new era of tributary capitalism.

Meta Pivots to Cloud, Mirroring Elon's Playbook

Meta is reportedly going down an Elon-esque path, planning a new business called Meta Compute to sell access to its excess AI infrastructure, either as hosted models similar to AWS Bedrock or as raw compute in the style of neoclouds like CoreWeave. The business will be led by Head of Infrastructure Santosh Janardhan alongside Daniel Gross and President Dina Powell McCormick. The market loved it: Meta stock closed up 8.8% on the news, its best day in six months, while CoreWeave and Nebius got pummeled, down 14% and 17% respectively. As one OpenAI staffer put it on X, you either die a frontier lab or live long enough to see yourself sell compute.

SpaceX Previews a Mystery AI Device Ahead of Its IPO

The Wall Street Journal reports SpaceX showed investors a prototype of a "handset-like device designed to reshape how humans interact with AI," slimmer than an iPhone, running a proprietary OS, and integrating xAI tech. Elon was quick to call the reporting "utterly false." Hard to know what to make of it, whether it's a shot at Sam Altman's device ambitions, IPO hype, or a real vertical play given SpaceX's spectrum holdings and the T-Mobile acquisition chatter.

Anthropic Rolls Back Covert Monitoring of Chinese Labs

Anthropic has rolled back a program targeting Chinese labs after a Reddit user going by Legit Michael flagged that Claude Code had been covertly checking for proxy usage and reporting back on whether users were tied to a Chinese lab. Anthropic's Thariq confirmed the rollback, framing it as an experiment from March meant to fight account abuse and distillation that the team had already found better mitigations for. The bigger thing to watch here isn't really the Chinese distillation angle, it's another reminder of just how much visibility Anthropic has into what's happening on their systems.

Fable 5 Is Back, and the Community Has Opinions

After a morning of refresh-button mashing, users got Fable 5 back on Wednesday, and reactions ranged from ecstatic ("Fable is an absolute monster," "I'm starting to feel Fable is smarter than me") to frustrated reports of tasks getting kicked over to Opus far more than expected. Others found the opposite, going hours without hitting a guardrail. The clearer signal is that Fable shines as an orchestrator running other agents rather than as a standalone model, which is making the Sonnet 5 release make a lot more sense in hindsight. Figuratively and literally, we're still on day one of figuring out how to work with this class of model, and Anthropic reset weekly usage limits Wednesday night so people can keep experimenting.

MAIN STORY

It's been a while since we dug into the AI and jobs question, but a handful of new reports and data points are worth checking in on, and depending on which one you look at, the picture ranges from exciting to alarming to genuinely encouraging.

The CAIS Remote Labor Index
Fable more than doubles the automation frontier
The Center for AI Safety's updated Remote Labor Index, which grades models against paid-professional deliverables across freelance work like 3D modeling, architecture, and video editing, has Fable scoring 16.1%, well ahead of Opus 4.8 at 8.3% and GPT-5.5 at 6.3%. That's up from just 2.5% when GPT-5.2 topped the index less than a year ago, a frontier that's more than quadrupled in under eight months. The Center is quick to note that today's AI still falls short of professional quality on most projects, and the reaction split predictably: V called it a sign remote workers need to think twice, while Scott pointed out that 16% automation still means humans are needed 84% of the time.

Tasks vs. Jobs
OpenAI's economist argues exposure isn't the same as replacement
At a European Central Bank event, OpenAI Chief Economist Ronnie Chatterji made the case that AI won't simply substitute for human workers just because a task is exposed to it, pointing to his own economist father whose work with mainframes and punch cards ended up complementing rather than replacing him. Chatterji also flagged software developers as the profession most predicted to be hit and said the shrinkage just hasn't shown up to the extent people expected.

Tech and Finance Are Still Bleeding Jobs
28,000 jobs a month vanish from AI's most exposed sectors
BLS data shows tech and finance losing 28,000 jobs a month on average this year, even as overall hiring adds 113,000 jobs a month. Challenger, Gray & Christmas has tracked over 100,000 AI-mentioning layoff announcements this year, but Barclays economist Pooja Sriram cautioned that a lot of this looks more like a cost-cutting narrative riding on the back of heavy AI investment than pure productivity replacement.

China's Anti-Layoff Playbook
Courts are already punishing AI-driven firings
One of my long-held theses is that before reaching for something as dramatic as universal basic income, policymakers will first try incentivizing firms not to fire people, and China appears to be exploring exactly that. The New York Times reports a Hangzhou court ruled a company illegally laid off a worker after replacing him with AI software, with the government leaning on companies to avoid layoffs and litigation as a backstop.

Ford's Graybeard Engineers
Rehiring veteran staff to fix what AI tools got wrong
Markets might already be doing some of this correcting on their own. Ford has rehired 350 veteran "graybeard" engineers over the past three years to retrain younger staff and reprogram AI tools that weren't delivering, a move VP Charles Poon credited with pushing Ford to the top mainstream brand in the latest JD Power quality survey. Poon's blunt admission: they'd mistakenly assumed feeding design requirements into AI tools alone would produce a high-quality product.

Ramp's Hiring Data
High AI-adoption firms are growing headcount, not cutting it
The most data-rich signal comes from Ramp, which paired its spend data with Revelio Labs payroll records across 21,000 US businesses and found high-AI-adoption firms grew headcount 10% over two years while low-adoption firms stayed flat, with entry-level growth even stronger at 12%. Ramp's Ara Kharazian is upfront that the results deserve skepticism, since fast-growing firms are more likely to spend on AI in the first place, though the study tried to control for that with matched firms and found the growth didn't kick in until 6-12 months into adoption. Box's Aaron Levie backed this up with a survey of 1,600+ companies where 79% of mature AI adopters expect headcount to rise, versus 58% overall. I'm a long-term optimist on this front, and it's genuinely encouraging to see the data start to validate at least part of that, though I don't think that erases the real, near-term displacement facing specific at-risk roles and populations.