AI is moving from chat and web search to delegated work.

Why it matters: The frontier AI labs have spent years promising that effective AI agents will act as our minions in the workplace and at home, and that might soon be a reality.

The big picture: Use of Codex — OpenAI's agentic coding and work platform — is accelerating, according to a new report from OpenAI, Columbia, Duke and the University of Pennsylvania.

Between the lines: The number of individuals using Codex is still small, but those who use it use it a lot, per the report, shared first with Axios.

Zoom in: Non-developers are the fastest-growing user group, even though software work is still the core use case for Codex.

Catch up quick: The shift to agentic work began in earnest at the beginning of 2026.

• That's when normal people began to allow Codex, OpenClaw, and Anthropic's Claude Code to interact with their desktops, manage calendars, read and write files, control web browsers, and execute scripts.

My thought bubble: As a journalist who has spent years covering cybersecurity — and whose coding knowledge tops out at early-2000s HTML — I was wary of giving agents access to my files, browser and apps.

• But over the last month, I've started using Codex and Claude Code for a lot of the work and life admin I used to handle manually.
• My agents fill out expense reports, triage email, make hair appointments and report packages stolen from my apartment lobby. Yes, this does happen often enough in San Francisco that I have saved time automating the task.

What they're saying: "Agents are reducing what I'd call the psychological cost of action," workplace culture expert Jessica Kriegel tells Axios.

• They "make unfamiliar work feel more approachable, which means I start sooner, experiment more, and spend less energy worrying about what I don't know."

Keep reading.

Facebook has "profoundly failed" in the AI race, former Facebook executive and venture capitalist Chamath Palihapitiya said on "The Axios Show."

The big picture: Palihapitiya, who helped Facebook grow into a global giant, told Axios' Dan Primack that the company "completely fumbled" what he viewed as a huge opportunity to lead in AI.

Flashback: Speaking at the Axios BFD summit in 2022, just before ChatGPT's public release, Palihapitiya said Meta was well positioned in AI because of the vast amount of context the company has about its users.

• In a Tuesday interview for "The Axios Show," Primack recalled that 2022 comment and asked Palihapitiya how the company fell behind.
• Palihapitiya declined to speculate about Meta's decision-making, saying he doesn't know enough about the company's internal dynamics.

Zoom in: He told "The Axios Show" that in the early days of ChatGPT and the growing chatbot mania, Facebook had the distribution and user base to immediately roll out AI products to a wide audience.

• Facebook had a chance to become the dominant champion of the open-weight AI ecosystem.
• Nvidia and CEO Jensen Huang better recognized the moment and built the infrastructure and ecosystem around open-weight AI, he argued.

Keep reading.

AI is consuming more and more AI-written content to formulate its answers — a feedback loop that could make its answers narrower, blander and easier to manipulate.

Why it matters: Regular search engines expose people to a variety of sources. If AI search comes to rely primarily on AI-generated content, it would shrink the range of information people use to form ideas.

The big picture: In simulations of AI search, models that relied on AI-generated reference material became increasingly likely to produce the same recommendations, according to new research from Graphite shared first with Axios.

• Graphite advises companies on how to improve their visibility in AI search.

Driving the news: Graphite's paper argues that AI search tools can experience "AI search collapse" when they retrieve AI-generated pages derived from earlier AI answers.

Catch up quick: Early consumer chatbots were often outdated because they answered mainly from training data, not live web results.

• When companies added web search to chatbots, the systems could retrieve current pages and use them to ground answers.

The intrigue: Graphite previously found that AI-generated content made up around half of all article-style web pages.

• If chatbots use those pages as source material, their answers could become less reflective of human taste, judgment and firsthand experience.

Zoom in: A study by Wharton researchers Gideon Nave, Christian Terwiesch and Lennart Meincke found that individuals who use ChatGPT as a research partner generated stronger ideas, but groups of people that use it tended to converge on similar concepts.

• Graphite's research suggests a similar narrowing could happen across AI search.

What they're saying: "If we're all using our brains less and just go to the LLM, we're all going to get the results that the LLM came up with," Nave tells Axios. "Then we're going to lose diversity and randomness and exploration."

Keep reading.

Former Commerce Secretary Gina Raimondo and former Indiana Gov. Eric Holcomb are launching Raise Us, a $500 million effort — with help from Anthropic and OpenAI's Foundation — to help states and employers prepare workers for an AI economy that could reshape jobs far beyond tech.

Why it matters: Competing AI labs are coming together to fund an effort aimed at addressing the labor market hit that their own technology could cause.

The big picture: Several tech companies have launched initiatives aimed at recruiting and training blue-collar workers who can facilitate data center builds.

• Raise Us focuses on AI's impact on the broader workforce, with a wider reach through public and private partnerships.
• "No single company or sector can solve the workforce challenges," Justin Spelhaug, president of Microsoft Elevate, said in a statement. David Zapolsky, Amazon's chief global affairs and legal officer, added this program is how you "make this transition work for everyone, not just a few."

Zoom in: Corporate participants — who will also fund the effort — include Amazon, IBM, Microsoft, Bank of America, Eli Lilly and more, as well as various state governments, educators and philanthropists.

• The group has already secured $500 million and hopes to raise $1 billion.
• The money will support programs including a startup accelerator for displaced workers learning how to start their own businesses or paid service years to give high school graduates a path toward AI-resilient careers.

Keep reading.

• Google could lose more staffers to rival AI lab Anthropic, after two high-level executives already left recently. (Bloomberg)
• The AI giants are all making the case that their water use is manageable. (Axios)
• Publishers of about 400 newspapers are suing OpenAI and Microsoft for using their content without permission. (Bloomberg Law)

Emily Sundberg, creator and author of Feed Me, a must-read Substack for the smart NYC girlies in my group chats, interviewed Mark Zuckerberg about the latest Kylie Jenner-designed Meta glasses.

Why it matters: Big Tech CEOs seem to be increasingly sold on the ROI of New Media instead of giving exclusive interviews to traditional news outlets.

The intrigue: Status, a media newsletter, reports that Meta had veto power over questions submitted by the audience.