Sam Altman now says AI has not wiped out as many entry-level white-collar jobs as he once feared, even if the long-term risk still hangs over the category. That shift matters because it pulls the labor story out of a straight-line apocalypse frame and back into a messier reality: model capability may be moving fast, but real organizations are still absorbing it unevenly, slowly, and with a lot more friction than the loudest narratives imply.

Uber is starting to say the quiet part out loud: token consumption and AI usage can rise sharply without creating a clean line to more useful product features shipped to users. That makes this a more important signal than it first appears. The enterprise AI story is moving from adoption theater into attribution pressure. At some point, leadership stops looking at usage dashboards and starts asking a much simpler question: what did all this spend actually buy?

China is restricting overseas travel for strategically important AI professionals at firms including Alibaba and DeepSeek, requiring approval for certain trips abroad. This is a strong signal that the AI race is no longer only about chips, capital, or datacenters. It is now reaching directly into human mobility. Once top talent is treated as a strategic asset, frontier competition moves one step deeper into state control.

Microsoft is opening a new front in the fight over data for AI agents | The Information frames Microsoft as fighting for the data, connector, and definition layers that agents need in order to act inside the enterprise. If that layer belongs to the platform, the moat in enterprise software can shift very quickly.

AI Can’t Agree on Which Jobs AI Might Destroy | Wall Street Journal brings the same question from Section 1 back to the center: how fast are models improving, and how far ahead of the evidence are corporate AI narratives about labor disruption? That makes it an important signal in the emerging labor-backlash cycle.

The future work may happen inside Codex or Claude Code, not inside SaaS UIs | Dan Shipper’s argument that instead of stuffing AI into every SaaS product, users may increasingly access those products from inside agent shells. If that happens, the primary software interface shifts from the app to the agent.

The harder B2B AI question: what is left if agent shells absorb all the value | Olivia Moore points to the downside of the new distribution layer. In the short term, being “available in ChatGPT / Codex / Claude” looks like a win. Over time, though, products that can be fully consumed inside those platforms risk being pushed down into little more than a data layer or context layer.

CEOs are vulnerable to “AI psychosis” because they do not live in the last mile | Levie puts his finger on a familiar distortion: executives often see the happy-path demo, not the verification, integration, cleanup, and edge-case work that real agent deployment creates. That is one reason executive optimism still runs ahead of operating reality.

Jordan Schachtel is pushing back against the UBI-for-AI storyline | Jordan Schachtel takes an openly opinionated shot at using AI job-displacement fears as the basis for a massive UBI program. It is not the strongest operator signal in the cluster, but it does show how quickly the labor story is spilling into open political combat.

Antigravity now has a much cheaper-token Gemini mode | Antigravity now offers Gemini 3.5 Flash (Low), with roughly 45% lower token consumption than the medium version. That is a clean sign that token efficiency is becoming a user-visible product feature, not just a backend optimization detail.

Muratcan Koylan’s post is useful because it turns a research-heavy idea like SkillOpt into a practical operating checklist. The core point is simple: in a self-editing loop, the thing that matters is not how many edits an agent can make. It is whether the validation gate is strict enough to reject weak ones. If your system accepts too many changes, you are probably shipping smarter-looking slop, not a better workflow.

If you want to apply this today, the first pass is straightforward:

The real payoff is that this moves agent optimization out of prompt craft by instinct and closer to an engineering discipline. For most teams that are not training their own models, that is the practical path for adapting frontier systems to their domain.

Tiny Scorecard:

That’s 6 sales in 7 months. 29 all time. And the performance?

16.5%, 17.6%, and 17.8%, net annualized returns on sold works held longer than one year (See all 29 at Masterworks.com)

It’s not from stocks, private equity, or real estate… it’s from contemporary and post war art. Crazy, right?

With Masterworks, you don’t need to be a BILLIONAIRE to invest in multi-million dollar art anymore.

Historically, the segment overall has had attractive appreciation and low correlation to stocks.*

Masterworks targets works featuring legends like Banksy, Basquiat, and Picasso, identifying what they believe to have significant long-term appreciation potential, not just at the artist level but at the level of individual artworks.

As one of the largest players in the art market, with $1.3 billion invested over 500 artworks, they pass critical advantages through to their 70,000+ members to add art to their portfolios strategically.

Looking to diversify your investments in 2026?

Click here to skip the waitlist

^{*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. See important Reg A disclosures at masterworks.com/cd.}

This paper matters because it brings the AI-and-jobs debate back to the ground through job-postings data instead of leaving it trapped inside CEO rhetoric or social-media fear cycles. If Section 1 today is about labor expectations softening, this is the deeper read that keeps that story anchored in something stronger: demand is shifting, but not in a single clean line.

Chris Olah offers a framing that is hard to ignore: frontier lab incentives can drift away from public-good outcomes unless something outside the industry pushes back. Put next to Uber and China in today’s issue, the piece works as a strong counterweight. This is not only a story about economics and geopolitics. It is also a story about legitimacy.