In partnership with
In Today’s Issue:
🥇 Anthropic passes OpenAI in Ramp's business adoption data
🤖 South Korean workers are training the next robot brains
🚧 Enterprise AI adoption crosses the halfway mark
🔍 OpenAI's safety record faces courtroom scrutiny
✨ And more AI goodness…
⚡ The Signal
Today's issue starts with a rare scene: an AI lab founder speaking at the Vatican about why AI is too important to be left to AI labs alone.
Anthropic co-founder Chris Olah used Pope Leo XIV's new encyclical to make a blunt point: frontier labs operate inside commercial, geopolitical, and personal incentives that can bend even sincere safety intentions. The bigger lesson is that AI governance is no longer just a technical debate. It is becoming a moral, institutional, and public-trust problem.
All the best,
Kim Isenberg
📊 Anthropic Passes OpenAI in Business Adoption
Ramp's May AI Index says Anthropic passed OpenAI in paid business adoption for the first time. Ramp reports Anthropic at 34.4% of businesses in April, OpenAI at 32.3%, and overall AI adoption at 50.6%, based on corporate card and invoice spend data.
👉 tl;dr: Enterprise AI is not settling into one permanent winner. Ramp's own caveat matters: this market is still fluid, cost-sensitive, and open to challengers that can undercut premium model pricing.
🤖 South Korean Workers Are Training Robot Brains
AP reports that South Korean startup RLWRLD is capturing skilled workers' motions to train AI software for robots. Lotte Hotel staff, CJ logistics workers, and Lawson convenience-store employees are being recorded so future systems can learn dexterous, real-world tasks rather than only repetitive factory motions.
👉 tl;dr: The next robotics race may be less about one flashy humanoid and more about who owns the best libraries of human motion, grip, timing, and workplace know-how.
🧬 FDA Starts Testing an AI Drug-Review Pathway
Intuition Labs' report says the FDA's Accelerated AI Pathway Pilot is aimed at AI-originated drug candidates entering early human trials. The pilot is framed around faster, more interactive IND review and clearer standards for evidence generated by AI in areas like target validation, dose prediction, and trial simulation.
👉 tl;dr: The exciting part is not simply faster drug review. It is regulators learning how to audit AI evidence before AI-designed medicines become ordinary clinical pipeline material.
When you read a big AI claim, ask your assistant to map the incentives around it before judging the claim itself.
Why it helps: Today's lead is not only about whether frontier models are powerful or safe. It is about who benefits, who pays, who has oversight, and who can say no when the incentives are pushing the other way.
Try this: Paste a launch post, policy speech, or research announcement and ask: "List the commercial, political, reputational, safety, and public-interest incentives shaping this claim. Then tell me which facts would be most important to verify independently."
🎬 Watch This
ABC News' Building A.I.: The Deep Mind Behind the Moment features Google DeepMind CEO Sir Demis Hassabis in conversation with Rebecca Jarvis about AI breakthroughs, regulation, energy, jobs, and what skills people should learn now. The useful contrast with today's lead is that Hassabis frames AI as a scientific accelerator, while still acknowledging the need for society, academia, and governance to shape the transition.
– Chris Olah, Anthropic co-founder, remarks at the Vatican presentation of Pope Leo XIV's AI encyclical
OpenAI's safety record is getting another airing in Elon Musk's lawsuit, according to TechCrunch. A former OpenAI AGI-readiness employee testified that the company became more product-focused over time, while former board member Tasha McCauley described weak confidence in the information the nonprofit board received before Sam Altman's 2023 firing and reinstatement.
The careful read: this is courtroom testimony in a live dispute, not a neutral audit of OpenAI. But the dustup keeps pointing at the same governance question: when frontier labs become huge commercial platforms, who can actually slow a launch or force a safety review?
AI Is Too Important to Be Governed From Inside the Lab
The Takeaway
👉 Pope Leo XIV released an AI encyclical, and Anthropic co-founder Chris Olah spoke at its Vatican presentation.
👉 Olah argued that frontier labs face commercial, geopolitical, and personal incentives that can conflict with doing the right thing.
👉 His core ask was for serious outside voices: religious communities, civil society, scholars, governments, and the public.
👉 The unusual part is the venue: AI governance is moving from technical policy rooms into moral and civic institutions.
The most important sentence in Chris Olah's Vatican remarks was not a model claim. It was an incentives claim. Speaking at the presentation of Pope Leo XIV's encyclical on AI, the Anthropic co-founder said frontier labs operate under pressures that can conflict with doing the right thing: commercial survival, the race to stay at the research frontier, geopolitics, pride, and ambition.
Olah's argument was more uncomfortable: even sincere builders need outside critics whose incentives are different. In his framing, AI raises questions that go beyond computer science because models are not engineered like bridges or airplanes; they are trained on human language and behavior, and even their creators still find parts of them mysterious.
The speech named three areas where broader discernment is needed: whether AI's gains will reach the global poor, what human flourishing should look like around widespread AI, and how society should think about the nature of increasingly complex models. You do not need to share the Vatican's theology to see the strategic point. The next phase of AI governance will not be won only by better benchmarks. It will need institutions outside the labs that can ask slower, more human questions.
Olah was very clear in his words about the dangers of the near future. Therefore, his words should be taken seriously.
“There is a real possibility that AI will displace human labor at very large scale. If that happens, supporting those displaced will be a moral imperative of historic proportions. This task will be difficult enough, but I worry most dialogue misses an even harder challenge. AI development is concentrated in a handful of wealthy nations. How can we ensure the gains of AI are shared globally? We do not have a mechanism for this. It is an unsolved problem, and it is the kind of problem the Church has historically refused to let the world ignore.”
Why it matters: If the people building frontier AI are also trapped inside the strongest incentives to ship it, outside scrutiny is not a nuisance; it is part of the safety system. The hard question is whether governments, civil society, and moral institutions can move fast enough to matter without becoming performative theater.
Sources:
🔗 https://www.anthropic.com/news/chris-olah-pope-leo-encyclical
Bloomberg: "No reliable safe havens." Billionaires have been investing elsewhere. Here's how to get in.
Bloomberg's Marcus Ashworth wrote plainly recently: "No more reliable safe havens."
After all, the S&P fell over 7% from the February peak. Bonds, even with less risk, are barely keeping pace with inflation.
So-called "diversified" portfolios have gotten hit from multiple directions.
Meanwhile, the world’s wealthiest have been setting records in another asset class.
Circumstances are always unique, but after the dot-com bust, it grew roughly 24% annually for a decade. After 2008, roughly 11% annually for 12 years.
Blue-chip art.
Why? It trades globally in multiple currencies, has scarce supply, and has shown near-zero correlation to equities since 1995.*
With Masterworks, 70,000+ investors allocated $1.3B fractionally across 500+ artworks featuring Banksy, Basquiat, and Picasso.
Accredited? You can invest in a diversified portfolio of postwar and contemporary art alongside two other real assets. From 2017-2025, the mix would’ve beat the S&P 500 by 3.1x.
See if you can improve your portfolio performance all in one diversified strategy.
*According to Masterworks data. Investing involves risk. Past performance is not indicative of future returns. Important Reg A disclosures: masterworks.com/cd
The chart: Ramp's AI Index shows the share of U.S. businesses with paid AI subscriptions rising from low single digits in early 2023 to just above half by 2026, with OpenAI, Anthropic, Google, xAI, and DeepSeek split out by vendor.
The lesson: Enterprise AI adoption is broadening, but the vendor race is still highly unstable; Anthropic's line accelerates sharply while OpenAI remains large.
The caveat: Ramp measures spend from its own corporate card and invoice data, so this is a strong business-spend signal, not a census of every company using AI.
Humanoid Robots Are Moving From Demo Videos to Factory Plans
⚡ Bottom line: Humanoid plans to deploy an estimated 1,000 to 2,000 robots across Schaeffler's global manufacturing sites by 2032, according to Reuters reporting carried by TimesLIVE.
💡 Why it matters: Humanoid robots are starting to look less like stage demos and more like phased industrial automation projects, with box handling and near-full-scale operations as the first practical targets.
🔎 What it means: Tuesday's robotics story is about the boring part becoming real: integration, actuators, rollout windows, factory lines, and whether humanoids can survive repetitive work outside a lab.
The humanoid robot story is shifting from viral videos to factory procurement. Reuters reports that British robotics company Humanoid plans to deploy an estimated 1,000 to 2,000 robots across Schaeffler's global manufacturing sites by 2032, with the first rollout scheduled for two German sites between December 2026 and June 2027.
The first use cases are deliberately unglamorous: box handling in Herzogenaurach and near-full-scale operational testing in Schweinfurt. That is exactly why this is important to understand. If humanoids become useful, the path probably starts with repetitive industrial tasks in places already designed around human movement.
The second half of the agreement may be just as important as the robots themselves. Schaeffler is also expected to become Humanoid's preferred supplier for more than half of its joint-actuator demand through 2031, covering a seven-digit number of components. In other words: the robot race is also a supply-chain race.
$20.8B in Redemption Requests. Percent Was Issuing Deals and Paying on Schedule.
Those requests came from non-traded BDC investors in Q1 2026, and most got back roughly half of what they asked for. Moody's U.S. BDC sector outlook: Negative.
On Percent's marketplace that same quarter: new issuances, scheduled payments, 0.44% lifetime net loss rate on asset-based deals since inception.† The difference is structural: concentrated corporate loans with redemption windows that close at manager discretion vs. asset-based finance with 6–24 month deal terms. 14.6% net ABS returns LTM after losses (3/31/26).† Starting at $500.
Alternative investments are speculative. No assurance can be given that investors will receive a return of their capital. †Past performance is not indicative of future results. Terms apply.
