✨ Rule of three | Axios AI+ Summary

1. Key Themes

Theme 1: AI Is Fracturing Society Into Three Distinct Camps — Not Converging

The mainstream narrative assumes AI adoption is a smooth, continuous curve. Instead, three hardened groups are forming — power users, doubters, and resisters — and the gap between them is widening, not closing.

"Three distinct camps are forming around AI: power users, doubters and resisters. AI isn't just advancing — it's fragmenting how people see the world."

"There is a growing gap in understanding of AI capability." — Andrej Karpathy

Theme 2: AI Adoption Creates a Self-Reinforcing Economic Advantage for Power Users

Those who go deep with AI tools don't just save time — they attempt harder tasks and succeed more often, creating a compounding productivity gap versus non-users. This is an investment signal: products and platforms that convert casual users into power users capture disproportionate value.

"Anthropic's March economic impact report found that experienced users attempt harder tasks and succeed more often. The result is a new kind of economic gap between advanced users and everyone else."

Theme 3: AI Resistance Is Escalating From Online Sentiment to Physical Violence

Anti-AI sentiment is moving offline. This is a material risk factor for AI companies — regulatory, reputational, and now physical security.

"In Indianapolis, a legislator said his home was hit by gunfire, with a note left behind saying 'no more data centers.' And on Friday, a man was arrested for allegedly throwing a Molotov cocktail at OpenAI CEO Sam Altman's home."

Sam Altman acknowledged: "The fear and anxiety about AI is justified; we are in the process of witnessing the largest change to society in a long time, and perhaps ever."

Theme 4: AI Companies Are Now High-Value Targets for State-Level Cyberattacks

The supply chain attack on OpenAI — linked by Google to a North Korean hacking group — signals that AI firms now face the same threat profile as critical national infrastructure.

"AI companies are now prime targets for classic software supply chain attacks — not just novel AI-specific threats."

"A GitHub workflow that the company uses to sign certificates for MacOS applications downloaded a malicious update from the Axios software on March 31... MacOS application users — including those for ChatGPT, Atlas and Codex — could have been affected."

Theme 5: Workplace AI Adoption Is Leader-Led, With a Growing Anxiety Gap on the Ground

Senior executives are using AI at dramatically higher rates than individual contributors, which means AI is being adopted top-down — creating a disconnect between those setting strategy and those fearing displacement.

"67% of leaders said they used AI daily or a few times a week, compared with 52% of managers, 50% of project managers and 46% of individual contributors."

"Among those who work in organizations that have adopted AI, [the share who believe their job could be replaced within five years] jumps to 23%."

2. Contrarian Perspectives

Contrarian 1: The Biggest AI Skeptics Are Often the Best-Informed — Not the Least-Informed

Conventional wisdom treats AI skeptics as people who simply don't understand the technology. The article explicitly distinguishes "doubters" (uninformed) from "resisters" (fully informed and opposed). The resister camp — including engineers and technical professionals — understand AI's trajectory and have actively chosen to oppose it.

"Resisters understand AI, think they know where it's headed and want no part of it." "A growing number of workers with technical skills fear AI will make them obsolete. In a viral post, a Meta engineer captured a spreading anxiety: 'I'm done with tech and I'm done with this unfair world.'"

This challenges the "education solves fear" approach many AI companies take — more information doesn't automatically produce more acceptance.

Contrarian 2: AI Adoption at Work Is Leader-Driven, But the Productivity Gap May Be Inverted

Most commentary focuses on frontline worker productivity gains from AI. But the Gallup data shows leaders are the heaviest users (67% daily/weekly), not workers. If AI primarily amplifies executive decision-making and strategy, the productivity gains may accrue at the top — not distributed across the workforce — intensifying, rather than reducing, organizational inequality.

"Differences by role 'reflect how many mainstream AI tools align with the tasks employees perform,' with leaders and managers taking on 'desk-based' responsibilities for which such tools can be 'readily applied.'"

Contrarian 3: A Single Free-Tier ChatGPT Session Is Defining Most People's Mental Model of AI Capability

The broad public perception of AI is anchored to its weakest, most accessible form — not its current frontier capabilities. This is a significant market inefficiency: companies and investors evaluating AI's potential based on casual public exposure are materially underestimating what power users are already doing.

Karpathy noted that "many people let a single session with ChatGPT's free tier define their view of AI" — while he personally spends "16 hours a day issuing commands to AI agent swarms and rushes to exhaust his tokens every month."

3. Companies Identified

OpenAI

Description: Leading AI lab, maker of ChatGPT
Why mentioned: Subject of a supply chain cyberattack linked to North Korean hackers; also referenced in the context of physical violence against CEO Sam Altman
Quote: "OpenAI has found evidence that one of its internal tools downloaded a compromised update from a recently infected, legitimate open-source software library."

Anthropic

Description: AI safety-focused AI lab, maker of Claude
Why mentioned: Their March 2026 economic impact report provides key data on the productivity gap between AI power users and casual users
Quote: "Anthropic's March economic impact report found that experienced users attempt harder tasks and succeed more often."

Box

Description: Enterprise cloud content management platform
Why mentioned: CEO Aaron Levie cited as an industry observer on the bifurcated nature of AI adoption
Quote: "AI adoption is a tale of two cities." — Aaron Levie

Apple

Description: Consumer technology giant
Why mentioned: Briefly noted as entering the AI glasses market, signaling hardware expansion of AI
Quote: "Apple has entered the AI glasses race, unveiling smart frames that bring artificial intelligence to your face."

Google

Description: Technology and AI giant
Why mentioned: Linked the OpenAI supply chain attack to a North Korean hacking group; CEO Sundar Pichai also cited on the US-China AI race
Quote: "Google has also linked the broader hacking campaign to a North Korean hacker group."

Arm

Description: Semiconductor IP company
Why mentioned: Newsletter sponsor; positioning its new AGI CPU for agentic AI workloads in data centers
Quote: "Arm is enabling this shift with the Arm AGI CPU, built for real-time orchestration at scale." (sponsored content)

Meta

Description: Social media and AI technology company
Why mentioned: A Meta engineer's viral post exemplified the growing anxiety among technical workers about AI-driven job displacement
Quote: "I'm done with tech and I'm done with this unfair world." — Meta engineer (viral post)

4. People Identified

Andrej Karpathy

Description: Former AI leader at OpenAI and Tesla; prominent AI researcher and commentator
Why mentioned: Cited as a canonical example of an AI power user and as a primary voice on the growing capability perception gap
Quotes: "There is a growing gap in understanding of AI capability." He also "spends 16 hours a day issuing commands to AI agent swarms and rushes to exhaust his tokens every month."

Aaron Levie

Description: CEO of Box, enterprise software leader
Why mentioned: Offered a sharp framing of the bifurcated AI adoption landscape
Quote: "AI adoption is a tale of two cities."

Sam Altman

Description: CEO of OpenAI
Why mentioned: His home was the target of a Molotov cocktail attack; his subsequent public response acknowledged the legitimacy of societal fear around AI
Quote: "It will not all go well. The fear and anxiety about AI is justified; we are in the process of witnessing the largest change to society in a long time, and perhaps ever."

Sundar Pichai

Description: CEO of Google
Why mentioned: Briefly cited as a voice on U.S. AI leadership in the global race against China
Quote: Referenced as saying "America needs to lead in the AI race." (via CBS)

5. Operating Insights

Insight 1: Convert Casual Users Into Power Users — That's Where the Value Compounds

The Anthropic data shows that experienced users don't just do the same tasks faster — they take on harder problems and succeed more. For product teams, the strategic priority isn't onboarding volume; it's depth of engagement. Invest in features, tutorials, and workflows that push users beyond basic prompting toward agent-level usage.

"Experienced users attempt harder tasks and succeed more often. The result is a new kind of economic gap between advanced users and everyone else."

Insight 2: AI Companies Must Treat Open-Source Dependencies as a Critical Security Surface

OpenAI's supply chain breach came not from a novel AI-specific attack, but from a compromised JavaScript library — a well-known attack vector that predates AI entirely. AI companies moving fast and integrating open-source tools broadly are inheriting significant, often unreviewed, security risk.

"AI companies are now prime targets for classic software supply chain attacks — not just novel AI-specific threats... A GitHub workflow that the company uses to sign certificates for MacOS applications downloaded a malicious update from the Axios software on March 31."

Insight 3: Address Worker Displacement Anxiety Directly — Ignoring It Creates Resisters

Organizations deploying AI without transparent communication about workforce impact are actively manufacturing the "resister" class. The data shows that exposure to AI adoption actually increases job displacement anxiety among employees — meaning rollout without change management backfires.

"Workers' concerns about their jobs being displaced have grown with AI adoption and organizational shifts... Among those who work in organizations that have adopted AI, [job replacement fears] jump to 23%."

6. Overlooked Insights

Overlooked Insight 1: Chinese AI Talent Is Leaving the U.S. — a Slow-Moving but Significant Shift

Buried in the "Training Data" links section is a Financial Times report that Chinese AI researchers are departing the U.S. due to tightening immigration policy. This is a potential long-term talent and geopolitical risk that could reshape where AI research concentrates — and which national ecosystems benefit — but receives almost no analysis in this issue.

"A number of Chinese AI researchers are leaving the U.S. amid tighter immigration policy and other concerns." (Financial Times, linked without elaboration)

Overlooked Insight 2: The PauseAI Movement Has an Organized Infrastructure

The article briefly notes that the suspect in the Altman attack "participated in a PauseAI Discord server" and "published anti-AI essays." PauseAI is described as "an activist group that advocates halting AI development." This is more than fringe online activity — it reflects an organized, ideologically coherent movement with communication infrastructure that warrants monitoring by AI companies and investors tracking regulatory and reputational risk.

"Someone with the same name as the suspect has published anti-AI essays and participated in a PauseAI Discord server. PauseAI is an activist group that advocates halting AI development."