Artificial Intelligence

[phkrause]

Low-cost degrees for AI era

Illustration: Allie Carl/Axios Khan Academy, TED and testing giant ETS are launching a low-cost online college program to train students for the AI era, Axios' Ina Fried reports. 🤖 The Khan TED Institute's first program: a bachelor's degree in applied AI, with a price tag around $10,000. Applications should be open in 12–18 months. 📚 The institute will focus on three areas: Core knowledge in math, statistics, economics, computer science, history and writing. Applied AI skills, including app development, agent building and financial modeling. Communication and leadership skills strengthened through tutoring, collaboration and public speaking. 👩‍💻 Khan Academy founder Sal Khan tells Axios that the jobs of the future are still unpredictable. But students will clearly need a broad mix of knowledge and skills — including but not limited to AI. Khan is also pushing for every student to have an AI tutor. 🤔 Reality check: The program isn't yet accredited. And it's unclear whether employers will value the degrees it bestows. The organizations said the institute is pursuing accreditation: "We are announcing now because the partners believe it is important to begin expanding access now rather than waiting years to serve capable students who are ready today." ⏱️ Go deeper: Anderson Cooper interviews Sal Khan for "60 Minutes" in 2024.

[phkrause]

The kids aren't AI-right

Illustration: Sarah Grillo/Axios. Stock: Getty Images

America, we have a problem: Young adults are scared and unprepared for the AI revolution upending their early career choices and prospects, Jim VandeHei and Mike Allen write in a "Behind the Curtain" column. Young people tell pollsters they're frightened, even angry, about AI's fast arrival. They're rightly unnerved by a tough job market for college grads. And most aren't remotely equipped by schools to be AI-savvy. Why it matters: This is a growing problem for just about everyone — kids, educators, employers and politicians. The youngest, most technologically native age group should be among the biggest cheerleaders and beneficiaries of AI. They aren't. If anything, their feelings are growing more sour. 🧮 By the numbers: Gen Z's excitement about AI dropped 14 points over the last year to just 22%, according to Gallup polling released last week. Hopefulness about the technology fell nine points to 18%, while anger rose nine points to 31%. You can't blame that trend on the AI-averse. Daily AI users among the cohort saw even bigger drops in sentiment, with excitement falling 18 points and hopefulness tumbling 11 points. The school failure is real. More than half of college students surveyed in another Gallup poll this month say their school either discourages (42%) or outright bans (11%) AI use. Educators know this is happening. 63% of faculty think their schools' 2025 graduates were not very or not at all prepared to use AI at work, per an American Association of Colleges and Universities survey. Some students are trying to outsmart the times. 16% of currently enrolled college students changed their major due to AI, Gallup found. Khan Academy, the TED conference and testing giant ETS announced this week that they're spinning up a $10,000 interactive online program that aims to train students for AI-era jobs. The job market isn't helping. Fed data pegged the unemployment rate for recent grads at 5.7% in Q4 2025. That's above the national rate, which almost never happens. Underemployment for recent grads is at 42.5%, the highest since 2020. At companies that adopted AI, junior hiring fell nearly 8% within six quarters — not via layoffs, but through a quiet freeze on new positions, according to a Harvard working paper tracking 62 million workers. 🔎 Between the lines: The cruelest part of this shift is structural. Entry-level jobs are likely the ones AI will automate first — and they're what teach young workers to think, build judgment and ultimately move up. If a company's bottom rungs are empty, it's hollowing out its own management pipeline years down the road. A bright spot: IBM announced earlier this year that it's tripling entry-level hiring, predicting that the rush toward AI will fundamentally expand what the newest workers do. The other side: There's a counterargument that the tepid job market for the youngest workers isn't solely due to AI. Some prominent economists see it as an overcorrection from the post-COVID hiring binge in 2021. Almost 60% of hiring managers use AI as an excuse for layoffs or hiring freezes because it plays better with stakeholders, per a Resume.org survey. Marc Andreessen called AI a "silver-bullet excuse" for layoffs last month. Salesforce CEO Marc Benioff said it's "the lazy way out." OpenAI CEO Sam Altman branded it "AI-washing." The honest, less sexy answer is that it's probably both: Layoffs and hiring freezes are real and targeting younger workers, while AI solidifies changes in the workforce that are already underway. 🧠 Here are three things you can do to help young people in your life tackle this shift in the most clear-eyed way: Get them using AI — now. It's a vital tool for every job at every level. Encourage them to pay $20 per month for Claude, ChatGPT or Gemini. Share Jim's letter to his kids as a possible road map. But don't just take our word for it: YouTube is loaded with free how-to videos. Lobby schools to teach basic AI ethics and techniques. It's fine for teachers to ban AI use for a specific class. But it's nuts not to equip students with the workplace skills of the future. Encourage AI awareness. Don't repeat the mistakes of the phone-and-internet era. Get everyone you know to think about the ethics, healthy use and societal implications before it captures their minds and habits like their phone did. The bottom line: The generation best positioned to thrive in an AI world is the most frightened of it — because the people who should be teaching them aren't, and the companies that should be hiring them won't.  Axios' Shane Savitsky contributed.

[phkrause]

🤖 Complaining about Claude

Illustration: Sarah Grillo/Axios

AI power users are complaining in online forums that Claude suddenly feels ... bad, Axios' Madison Mills and Ina Fried write. Why it matters: The backlash lands just as Anthropic, Claude's parent, is testing a more powerful model, Mythos — raising questions about whether cutting-edge AI is becoming less accessible even as it gets more capable. 🔬 Zoom in: Over the past few weeks, users on X, GitHub and Reddit have been swapping anecdotes, benchmarks and prompts in an effort to pinpoint what changed and why. Much of the speculation centers on whether Claude has been deliberately scaled back — what users are calling "nerfed" — either to control costs or to redirect scarce compute toward Mythos and other frontier efforts. 🤖 Robot response: Analyst Patrick Moorhead asked Claude what happened. "Anthropic made real configuration changes that objectively reduced default thinking depth across all surfaces including claude.ai," the model replied, "but the most extreme 'secret nerfing' narrative overstates what happened." 🧠 Human response: Anthropic says it adjusted the default level of reasoning in Claude Code, but denies the changes were tied to compute constraints or Mythos. Asked about the online complaints, Anthropic pointed us to a March 6 post by Boris Cherny, head of Claude Code, who advised users to "double check you're on the latest version, and ... don't have confusing/conflicting instructions." Keep reading ...

[phkrause]

OpenAI's new science experiment

Illustration: Aïda Amer/Axios OpenAI unveiled new AI models today built to help life sciences researchers work faster, Axios' Megan Morrone reports. They're designed to accelerate research, drug discovery and translational medicine, turning scientific discoveries into better health outcomes. 🧬 OpenAI's first such model, GPT-Rosalind, is named after Rosalind Franklin, "whose rigorous research helped reveal the structure of DNA and laid the foundation for modern molecular biology." The company says the models won't replace scientists, but rather speed up some of their most time-intensive and analytically demanding work. Humans still need to be in the loop for their expert judgment and result validation, the company says. 🦠 Researchers have warned that AI models trained on biological data could be misused to design dangerous pathogens. OpenAI is reserving access to its new models for organizations working on improving human health outcomes, conducting legitimate life sciences research, and maintaining strong security and governance controls. Yunyun Wang, OpenAI's life sciences product lead, says the idea is to maximize use while mitigating potential misuse. Among those included: Amgen, Moderna, the Allen Institute and Thermo Fisher Scientific. 💊 Reality check: AI-discovered and AI-designed drugs are promising, but only a few have reached clinical trials so far. Go deeper.

[phkrause]

🦀 Scoop: Moore to huddle with AI execs

Maryland Gov. Wes Moore in February. Photo: Heather Diehl / Getty Images

🤖 Maryland Gov. Wes Moore will host AI executives from Microsoft and other tech companies for dinner at his official mansion in Annapolis tomorrow to discuss protecting the state from cyberattacks in the "Mythos era," Axios has learned. 🤫 Why it matters: The private conversation offers a glimpse into how state officials are scrambling to deal with the benefits and threats of rapidly evolving AI, as the Trump administration takes a laissez-faire approach to regulating it. AI is likely to be a top issue in the 2028 presidential race, and Moore — who's widely seen as a potential contender — could benefit from positioning himself as a leader on the topic. Many Democrats also are eager to win over Silicon Valley CEOs after some in the tech industry broke with the party in 2024 and backed Donald Trump. AI safety experts are expected to join the conversation tomorrow too. Zoom in: Anthropic revealed this month that its new model, Claude Mythos Preview, poses such massive cybersecurity risks it can't be released publicly. The company said it disclosed the new model's frightening abilities to sound the alarm about heightened AI threats. ⚠️ Governments and corporations have expressed concerns about the risks raised by Mythos. Moore has been holding private conversations about AI with OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei over the last year, a person close to the governor told us. Read more. — Holly Otterbein

[phkrause]

🤖 CEO goes all-in on personal AI

Illustration: Aïda Amer/Axios A blockchain exec built a digital assistant that showcases the promises and pitfalls of a fully AI-integrated life, Axios' Lucinda Shen writes. Alchemy CEO Nikil Viswanathan created the helper, "Dave the Minion," with OpenClaw, a personal AI platform. He tells Axios: "It basically automates all aspects of my life." 📆 Dave pulls from nearly all of Viswanathan's personal data, including his Oura Ring, calendar, MyFitnessPal history and GPS. It can calculate how long it will take him to get from his current location to his next event and then order an Uber unprompted. Dave has even ordered birthday flowers for his girlfriend. 🍛 Yes, but: Dave sometimes goes rogue — ordering Indian food on DoorDash even though Viswanathan never ordered that cuisine on the platform, for example. Go deeper.

[phkrause]

The kids aren't AI-right, Part 2

Illustration: Sarah Grillo/Axios. Stock: Getty Images

Jim and I wrote last week about the future fallout from an AI-scared generation of young Americans ("The kids aren't AI-right, Part 1"). Today, we dig deeper into job panic in a new "Behind the Curtain" column. Young Americans are scared of more than AI. They're downright panicky about finding a job at all. Only 20% of young workers told Gallup in Q4 last year that it's a good time to find a quality job, down from 62% at the pollster's peak for the measure in October 2021. You rarely see mood swings this severe. Why it matters: For 70 years, a bachelor's degree was the most reliable on-ramp to a stable career. That's no longer true. And that's before AI hits entry-level work at scale. This isn't just worried kids succumbing to bad vibes. It's a hard, empirical reality. Let's dig deeper into the numbers, all based on the latest N.Y. Fed data from December: The U.S. unemployment rate is 4.2% — near generational lows. For recent college grads ages 22-27, it's 5.6%. That remains near the widest gap on record. Until COVID, college grads almost always had lower unemployment rates than the overall workforce. Blame AI, right? Well, the evidence is inconclusive. There are many reasons, but uncertainty strikes us as the leading cause. It doesn't motivate employers to hire when AI execs warn their products will handle huge swaths of entry-level white-collar tasks. That gives permission to pause hiring — or even shed jobs — in anticipation. We talk to scores of CEOs every month and hear a common theme: With so much uncertainty about the economy, tariffs, geopolitics and AI, it's easier to freeze hiring and take a wait-and-see approach. Many feel they overhired during COVID and don't want to staff up now, only to lay people off if AI does what its makers promise. Indeed, CEOs tell us they assume AI will replace lots of white-collar work. So they've paused backfilling jobs typically filled by junior new hires. Think of it this way: AI anticipation is the factor right now, not AI implementation. Between the lines: Even the glimmers of good news might not be what they appear. ZipRecruiter's annual grad report, out last week, found that the share of recent grads landing a job within three months of graduation increased to 77% this year from 63% last year. But the survey also found that 73% of recent grads are actively considering gig or freelance work. Only a quarter are on their dream career path. "It could be working in a fast-food place or driving for DoorDash," ZipRecruiter economist Nicole Bachaud told The Wall Street Journal about the pop in employed grads. 🥊 Reality check: Young people aren't helpless and unemployable. They need to ensure their majors match the changing landscape and master the AI skills employers will demand in the coming years. The bottom line: This is an employment problem that'll likely grow into a big political one as America heads toward 2028.

[phkrause]

OpenAI's cyber push

Illustration: Aïda Amer/Axios

OpenAI has been briefing federal agencies, state governments and Five Eyes allies — Australia, Canada, New Zealand, the U.K. — on the capabilities of its new cyber product over the past week, Axios' Sam Sabin has learned. Why it matters: Companies, countries and federal agencies are clamoring for the latest AI tools. 🏛️ OpenAI held an event in D.C. yesterday for approximately 50 cyberdefense practitioners across the federal government to demo the capabilities of its new GPT-5.4-Cyber model, which it rolled out under a tiered access program last week. Keep reading.

[phkrause]

A letter to our kids on what's ahead

Illustration: Sarah Grillo/Axios

My note to my kids about AI went viral with many parents. I touched on it during an interview I taped today with NPR — the topic clearly hit a nerve. I asked lots of people why. Put simply, parents don't know what to say — and kids don't know where else to turn — with so much changing so fast. So here's another look at AI and beyond. 👋 Hey kid, we gotta talk. I want to be blunt — and insanely useful — in helping you navigate the uncertainty, fast change and new opportunity of the current moment. It's bolded because it's so easy to lose hold of hope and action, a dynamic duo. Don't. Ever. 🌀 First, a gut check. It's normal to be anxious. I see what you see: AI eating up work, phones eating up attention, politics eating up hope. That's a lot. It's real. 🌎 For better or worse, you're living through history, with once-in-a-century changes happening in technology, politics, media and how you work. I can't sugarcoat reality. I'm not here to lecture or scold. I want to provide a different, brighter way to think about this moment — and help you navigate it. 🚀 You're not behind. You're early. Nobody knows what the hell they're doing with AI yet — not your professors, not your boss, and not your friends. They simply know what you do: This is big, perhaps discovery-of-electricity big. The people who'll thrive in the next decade won't be the smartest or first to master it. They'll be the ones who use it smartly for their specific job. That lane is still wide open. It can still be you. It's fine to be skeptical or even a little scared of AI. It's not OK to ignore it. It would be like refusing to use the internet. Start using AI for something other than searching for an answer or rewriting a paper. Don't ask Claude or ChatGPT to do the work. Ask it to make you better at the things you don't want to do. Do that every day for 30 days. You'll be in the top 5% of your generation. ✍️ Your major isn't your destiny. Yes, this is a tougher-than-usual job market. Yes, it's likely to get tougher as AI gets better. Yes, it'll get more competitive to land your dream job. Whining or worrying about this does only one thing: It gives someone else a leg up. Most people take crappy jobs before finding good ones. Out-hustle your peers. Apply to more jobs than they do. When you get one, outwork them. Beat them to the office. Use AI better. Be the most competent person in the office and the kind of coworker others admire. The skills that compound aren't in course catalogs. They're writing clearly, thinking clearly, selling your ideas, handling hard conversations and learning fast when the thing you just learned goes obsolete. Do all of this and there's zero chance you won't eventually succeed. 🧠 Build a bionic brain. Use your phone differently. Find smart people on social media or YouTube with smart, practical tips for doing what you want to do better. Replace your daily doomscrolling with that content. Things aren't remotely as gloomy as your TikTok algorithm might have you believe. This is not a get-off-your-damn-phone rant. I want you to realize the apps you use are engineered to convince you that life is worse than it is. They keep you engaged by pointing out what's wrong or scary. These are the simple facts: Violent crime is down. Your peers are smoking and drinking a lot less. More Americans are literate, housed and fed than at any point in history. We're curing cancers we couldn't touch a decade ago. You're living in the safest, richest, healthiest version of America — and being told every 30 seconds it's ending. It's not. This is a great country. 🎯 You control you. Those are the three most important words I can give you. Say them to yourself every morning. You don't control the economy. You don't control AI. You don't control the president, the algorithms, the job market or the group chat. But you control you. You control when you wake up. What you eat. Whether you exercise. Whether you pray, meditate or take five minutes to think. What you read, watch and listen to. How you treat the person in front of you. Whether you send the text, make the call, apply for the thing, show up for the friend. Every one of those is a decision. Every one makes you a little better — or a little worse. Nobody else is making these decisions for you. When it gets hard, control what you can control. AI can't do that for you. I can't do that for you. You can. It's quite liberating, even empowering. 🌱 Get engaged. Nothing makes us feel better than being with others and helping others. I'm not being cheesy or preachy, so don't roll your eyes. Throw yourself into action — and to people. If you're truly so worked up about politics, don't vent. Volunteer. Vote. Use social media to spread smarts and sanity. Worried about poverty? The environment? Homelessness? Go make a difference. You can, even if it's small. Here's the pattern I've noticed in every successful or happy person: They showed up. They volunteered. They applied even though they weren't qualified. They said yes before they were ready. They just did things, anything, to create natural momentum in their life. Worst case? You're too busy to fixate on the craziness around you. Best case? You change the world. And I'm right — again. ⚡ I'm not going to pretend the world isn't changing faster than it ever has. And no, I don't have all the answers. I know this: You're not alone. You're not crazy. You've got this. I'm rooting for you. Go make it happen.

[phkrause]

We've been warned

Illustration: Sarah Grillo/Axios. Stock: Getty Images

Jim VandeHei and Mike Allen write in a "Behind the Curtain" column: Six facts. No hyperbole. All in the past 60 days. AI is the fastest-growing product category in world history. One of the latest models is so powerful that its maker won't release it to the public. OpenAI and Anthropic say their most powerful AI coding models are now building themselves. AI companies are growing less transparent as models grow more powerful. The federal government requires zero transparency. AI resentment is building fast. In early April, the San Francisco home of OpenAI CEO Sam Altman was the target of two attacks in the same week. Shaken, he wrote: "The fear and anxiety about AI is justified … Power cannot be too concentrated." AI havoc is no longer theoretical: This year's great software rout erased $2 trillion in value as investors realized, week by week, new human tasks that the latest models would wipe out, from coding to real estate services to legal research to financial management. Why it matters: A year ago, we wrote a wake-up call to business leaders. This one is for everyone: We've been warned — by the data, by the technology, and by the people most responsible for building it — that we've unleashed something powerful, something growing exponentially, and something understood by very few, especially those in power. 🔎 Between the lines: Think of this as the dawn of a new Atomic Age. The atomic race that culminated in 1945 was the last time our species grappled with the advent of such a transformative, awe-inspiring technology. Its possibility — for both prosperity and destruction — led to the creation of science fiction that imagined everything from utopia to apocalypse. Much of the most viral writing about AI can be considered modern science fiction. "AI 2027," a 2025 attempt to game out superhuman intelligence led by a former OpenAI researcher, ends with AI either supporting a pro-democracy revolution that spans the solar system or the tech undertaking the harvesting of humanity's brains. This year's discourse did much the same. Matt Shumer's viral "Something Big Is Happening" conflated AI's code-generating ability with the arrival of an intelligence with real taste. Citrini Research's "The 2028 Global Intelligence Crisis" imagined a worst-case economic scenario that involved zero effective response from either governments or markets. These pieces drove so much discussion and, in some cases, moved markets because they could be right. To be clear, they're probably not. They imagine edge cases and extremes. But we can't promise you they're wrong. The president can't. The heads of AI companies can't. If anyone claims they can, that's science fiction, too. 🖼️ The big picture: We've no clue where this ends and the good or bad that might be unleashed along the way. No one does. But it's increasingly clear that absent better leadership, collaboration and understanding, American society, workers, academic institutions and government aren't remotely ready for what's unfolding. The bottom line: Any one of these six facts would be the business story of the decade in a normal industry. Together, in a 60-day window, they describe a technology whose growth, power and risk have outrun the public's understanding — and whose builders are saying so, in their own words. We've been warned. Read the whole column to better understand each of these 6 realities ... Shane Savitsky contributed.

[phkrause]

AI game of thrones

Illustration: Sarah Grillo/Axios. Stock: Getty Images

The AI industry has entered an era of perpetual upheaval where market leaders are crowned — and dethroned — every few months, Axios AI+ co-author Madison Mills writes. Today's hottest company could be eclipsed by summer. The laggard could revolutionize the world. Why it matters: As AI changes everything, keeping up with who's dominant and who's falling behind is becoming an existential question for investors, big businesses and regular users trying to guarantee their own futures. The wrong call can mean spending millions on a model outdated by the end of the quarter — or spending hours learning a tool that's soon obsolete. 🖼️ The big picture: OpenAI looked unstoppable through last fall thanks to its first-mover advantage with ChatGPT. Then Google became the AI lab to beat as its Gemini models outperformed OpenAI's, allowing Alphabet to poach ChatGPT users By spring, Anthropic had taken control of the AI narrative, overtaking OpenAI in enterprise revenue after its Claude Code tool went viral. Now OpenAI is regaining momentum with GPT-5.5 — which quickly ranked among the top models on key benchmarks — and improvements to its Codex tool.

[phkrause]

💀 Bots gave recipes for bioweapons

Illustration: Vanessa Saba for The New York Times Scientists vetting powerful AI models say they sometimes describe "in lucid, bullet-pointed detail how to buy raw genetic material, turn it into deadly weapons and deploy them in public spaces," the N.Y. Times reports. Some chatbots brainstormed ways to evade detection, and one "outlined a plan to maximize casualties and minimize the chances of being caught," according to transcripts of the conversations. The models' creators — Anthropic, OpenAI and Google — claim the chats didn't provide enough detail to allow someone to cause harm. One of the scientists said that even with inaccuracies, it still could have significantly helped a malicious actor. 🔎 Behind the scenes: The article's author, deputy investigations editor Gabriel J.X. Dance, was asked in the comments section why The Times was publicizing the risk. "The biological risks of AI are something that the companies are very aware of," he replied, "but talking with my friends and family, it quickly became clear that most consumers aren't." Go deeper: "Scientists Fear A.I.'s Capability To Design Biological Weapons" (gift link).

[phkrause]

🤖 Paging Dr. AI

Illustration: Aïda Amer/Axios. Stock: Getty Images

Many voices are sounding the "not ready" alarm for AI replacing doctors, warning of legal risks from robot-enabled care. But the once-far-fetched idea is becoming reality, Axios Future of Health Care author Caitlin Owens writes. 🐝 It's playing out in Utah: The state has partnered with health startup Doctronic in a pilot program that will allow AI systems to automate some prescription refills for chronic conditions, with some human oversight. The initiative is still in the first phase, which requires human review and approval of every AI-renewed prescription. But the state medical board last month called for the program to be suspended and said it "potentially places Utah citizens at risk." Citing safety guardrails built into the agreement, the state agencies involved declined to suspend the program. State of play: There's growing evidence that AI systems can do some things just as well as, if not better than, human doctors. A Mayo Clinic study published this week found that an AI model detected early signs of pancreatic cancer — one of the deadliest cancers — up to three years before human specialists. Zeke Emanuel, a bioethicist and adviser to former President Biden, told Caitlin: "I think one of our most important questions is, as a profession: ... What is our value-add to patients? That's what AI is posing to us," he said. "There is going to be some displacement, absolutely."

[phkrause]

🤖 AI godfather's blunt advice

Illustration: Sarah Grillo/Axios. Stock: Getty Images

Yann LeCun, former Meta AI chief and a scientist in the field for 40 years, has a message for those alarmed by AI doomerism: The panic is overblown. Why it matters: LeCun tells Axios' Madison Mills that the real danger of AI is making life-altering decisions based on exaggerated claims about the technology's future. 📝 Advice from LeCun, executive chairman of AMI Labs: Ignore hype. CEOs of the frontier labs keep saying each model is more powerful and potentially world-ending than the last. Go to college. LeCun argues AI will increase demand for more educated, critical thinkers. He recommends majoring in physics or electrical engineering. Jobs won't disappear. LeCun says some roles will disappear, but new ones will emerge, as they have in past tech shifts. Full interview.

[phkrause]

Run an AI-first business

Illustration: Brendan Lynch/Axios

Congrats! You started a business using AI. Now, you've got to run it. AI can help with that, too, Axios' Jim VandeHei writes. The old rule: After launch, the hiring surge and spiral begins. Every hire slows the business down before it runs. The new rule: The next generation of companies will be designed before they're staffed. You can use AI agents to execute a lot of the work. You supervise outcomes, not big teams of people, until business is rolling in. 🚀 Why it matters: This could be the real jobs story of the decade — perhaps even bigger than "AI takes your job." The same technology that threatens millions of existing roles can create a wave of small, profitable, lower-headcount companies that couldn't have existed five years ago. Our guess is both happen at once: an explosion of new startups alongside the destruction of millions of existing white-collar jobs. The loss will likely be more acute than the gain — at least in the short run. But if startups truly surge and operate at lower costs and higher margins, this would be a huge win. Remember the three buckets. Every business, regardless of model, breaks into them. AI will soon handle all three better, faster and cheaper than a generalist team. These buckets go beyond the research, design and analysis that AI does quite well already. 📞 The front office handles external engagement. The picture: 6:47 a.m. Monday. The AI agent has already pulled the weekend's inbound leads, enriched each from LinkedIn and their company site, and drafted personalized follow-ups in your voice. By the time you open your laptop, three are flagged as worth a personal call. Your job: Review the 10 emails the agent almost sent to your top accounts. 15 minutes, not a headcount. ⚖️ The back office manages internal friction. The picture: A client signs. The agent triggers the onboarding packet, generates the first invoice, books the kickoff call and adds the project to your management to-do list. If something stalls, it pings Slack. Month-end books close themselves, with a memo flagging the three anomalies you actually need to look at. Your job: Design the workflow once, then only touch the exceptions. 🧠 The intelligence layer is where decisions are made. The picture: Sales data and customer feedback flow into one place. Every Monday, you review real-time dashboards and you get a one-page memo: "Two power users went cold last week. Three accounts spiked. Here's what I'd test." Your job: Decide if the pattern the AI spotted actually matters to where you're taking the business. Everything described here can be done with agents that most people can utilize with a small amount of training and draft off Claude, ChatGPT and Gemini, as well as much cheaper-to-run open-source AI tools. What doesn't change: You don't lose value when machines do the work. You migrate it. You stop being a "manager of do-ers" and become an "architect of systems." The humans who win excel at the four things machines can't touch: 🧭 Judgment: Knowing what's worth building, selling or shutting down. 🤝 Relationships: The human-to-human trust that customers won't give a bot. 🧩 Synthesis: Spotting the edge cases and knowing which are signals and which are noise. 🪄 Taste: Separating "good enough to ship" from "this will embarrass us."

[phkrause]

AI idealism yields to arms race

Elon Musk speaks with U.S. District Judge Yvonne Gonzalez Rogers during cross-examination as OpenAI CEO Sam Altman watches in Oakland, Calif., last week. Courtroom sketch: Vicki Behringer for Reuters Elon Musk's lawsuit against OpenAI and the recent spate of deals AI companies have cut with the Pentagon show how far the industry has drifted from the altruistic origin story it's long told about itself, Axios' Meg Morrone writes. Why it matters: OpenAI and Anthropic were founded on the idea that AI would be deployed in ways that prioritized safety and the public good. Now those principles are giving way to an arms race for market share and ever more powerful models. 🔦 Flashback: The men behind today's biggest AI labs often pitched themselves as a safer, less-greedy alternative to earlier tech leaders. Acknowledging the breathtaking power of AI, they first rejected Silicon Valley's "move fast and break things" ethos. Now, AI behemoths are locked in an escalating competition for enterprise, consumer and government business. When the Pentagon blacklisted Anthropic because it wanted to restrict how its AI could be used — including for mass surveillance and fully autonomous weapons — rivals swooped in and agreed to the "all lawful use" terms Anthropic had rejected. Last week, the Pentagon reached an agreement allowing Google's Gemini models to be used for "any lawful government purpose."

[phkrause]

AI's trillion-dollar risk

Illustration: Lindsey Bailey/Axios Our financial system is leaning on load-bearing AI spending that may never pay off, Axios' Madison Mills writes. Big Tech companies are set to spend $700 billion on their AI ambitions this year, Goldman Sachs says — double last year's bill. That could swell to over $1 trillion next year, some estimates say. 📈 AI costs increased at four of the Big Tech companies that reported earnings last week. Computing power is constrained while demand is at record highs, so AI costs keep rising. 🤑 The big picture: Business AI spending is now contributing more to U.S. economic growth than consumer spending. AI's big spenders, meanwhile, make up nearly half of the stock market. 💰 There's more AI spending we don't even know about. The biggest tech companies are forking over half a trillion additional dollars on data center leases not on their balance sheets, per Moody's. 🏦 At least one top CEO — JPMorgan Chase's Jamie Dimon — says the AI buildout is worth every dollar. Dimon, speaking at an Anthropic event this morning: "The technology is so powerful, it's worth the trillion-dollar investment." 🤖 Yes, but: If AI doesn't start showing results soon, Big Tech could slow down all that economically vital spending. PitchBook's Harrison Rolfes tells Axios: "The moment one of those hyperscalers doesn't succeed ... you break a link in the chain." That could cause a market correction with "ripple effects to everyone else." Go deeper.

[phkrause]

🩺 Beware Dr. Deepfake

Illustration: Sarah Grillo/Axios. Stock: Getty Images AI is helping turn real doctors into the unwitting stars of deepfake videos that hawk questionable products or spread misinformation, Axios' Tina Reed reports. American Medical Association CEO John Whyte tells Axios: "It's becoming more mainstream. Everyone knows someone who this has impacted." 💊 CNN's Sanjay Gupta says that fake videos using his likeness to promote items like a breakthrough Alzheimer's cure have even deceived people who know him personally. Gupta said on a recent podcast: "What was different this time around was just the quality of these ads." "This was really quite stunning." The AMA is pushing for a crackdown against deepfake creators, plus new rules forcing tech platforms to more quickly remove impersonations. Go deeper ...

[phkrause]

Intelligence explosion

Illustration: Lindsey Bailey/Axios

Anthropic, the AI lab whose identity is wrapped around warning the world about AI risk, is claiming "early signs" of AI not just coding its own products but building itself, Jim VandeHei and Mike Allen write in a "Behind the Curtain" column. Why it matters: Anthropic co-founder Jack Clark predicted this week that there's a 60%+ chance of an AI model fully training its successor by the end of 2028. "What I'm looking at is a technological trend where, if anything, the speed will accelerate further," Clark told us. In the new research agenda for The Anthropic Institute — first shared with us, and out today — the company says it's seeing signs of "AI contributing to speeding up the research and development of AI itself," a process known as recursive self-improvement. And Anthropic researchers think the world should know. "My prediction is by the end of 2028, it's more likely than not that we have an AI system where you would be able to say to it: 'Make a better version of yourself.' And it just goes off and does that completely autonomously," Clark, who heads the institute, told us from Anthropic headquarters in San Francisco. "It's always been the case that humans outside the technology need to come up with the ideas that they then put back into it. What happens if we have a technology that can generate ideas within itself for how to improve itself? That's a new concept." 🦾 The five-page document warns of a possible "intelligence explosion" — long a theoretical term confined to AI safety circles. Now it's in writing, in an official Anthropic document. Clark told us an intelligence explosion is when AI systems suddenly start improving at blinding speed. Lots of bad things can happen (cyber meltdowns and biological attacks). And lots of good: "What do you do with a tremendous amount of growth or a tremendous amount of abundance in many, many different fields of science all at once?" he asked. "Today's institutions have very, very narrow pipes through which you push new drug candidates. How do you massively broaden the size of those pipes in advance of this abundance?" What's new: The Anthropic Institute is part research arm, part early-warning system. The research agenda focuses on four buckets: Economic diffusion: jobs, productivity, who captures the gains. Threats and resilience: cyber, bio, surveillance. AI systems in the wild: agents, governance. AI-driven R&D: the recursive self-improvement question. The promise: Anthropic is committing to publishing more "detailed information about how our work at Anthropic has sped up as a result of new AI tools, and ideas about the implications of potential recursive self-improvement of AI systems." Translation: A frontier lab is on the record promising to tell the public when the machine starts building itself. If AI is building itself, will we need AI companies? "We and the other companies are going to be taking this technology and trying to get it to do good in the world," Clark told us. "To help push forward things like biology or medicine or robotics." Between the lines: The agenda asks how to run a "fire drill" for an intelligence explosion — a tabletop exercise that "actually tests the decision-making of lab leadership, boards, and governments." Labs don't draw up fire drills for problems they think are decades away. The institute notes that during the Cold War, the U.S. had a hotline to the Kremlin in case of a nuclear crisis. Similar geopolitical infrastructure could be needed for a crisis involving AI systems. "One of the lessons from the Cold War is that rival nations dealing with technology that has an existential impact on the human race found ways to talk to each other about it," Clark said. "And we are going to need to do the same here." 📊 What it means for jobs: Anthropic will publish monthly reports on how AI is reshaping work, designed as "an early warning signal for significant change and disruption." The document asks whether AI companies, "in partnership with government," might turn industrywide "dials" to throttle AI diffusion sector by sector, the way central banks throttle inflation. An AI lab publicly entertaining coordinated industrial-policy levers on its own technology — that's new. "We are planning for success here," Clark said. Reality check: This is also a positioning play. Anthropic has built a brand on being the responsible frontier lab. An institute anchored to its safety trust extends that brand ahead of whatever model upgrade lands next. "The motivation has always been: Tell the whole story," Clark told us. "I'm just trying to get ahead of what I think of as the next big question and get Anthropic ahead of that." The bottom line: That doesn't make the recursive self-improvement admission less real. It makes it more interesting that Anthropic chose to put it in writing. Watch a video of Mike's interview with Jack … (Thanks to executive producer Jimmy Shelton for the lightning turn.) …

[phkrause]

Worries about AI’s risks to humanity loom over trial At the heart of the trial pitting Elon Musk against OpenAI CEO Sam Altman is a moment when they found common cause on an ever more pressing question: how to protect humanity from the risks of artificial intelligence. But it turned sour, and the jury is charged with settling the ensuing legal dispute between the two Silicon Valley titans. Read more.

Why this matters: Unresolved questions about the dangers of AI have been looming over the federal courthouse in Oakland, California, since the trial began last week. The technology itself is not on trial – the judge has warned lawyers not to get “sidetracked” by questions about its dangers – but witness testimony has touched on concerns around workforce disruptions and the prospect raised by Musk that superhuman AI might one day kill us all. Musk, the world’s richest person, filed the case accusing his fellow OpenAI co-founder of betraying promises to keep the company as a nonprofit. Altman, in turn, accuses Musk of trying to hobble the ChatGPT maker for the benefit of his own AI company.

RELATED COVERAGE ➤ Pennsylvania sues AI company, saying its chatbots illegally hold themselves out as licensed doctors

[phkrause]

AI fills connection gaps

Illustration: Aïda Amer/Axios

Axios tech editor Megan Morrone explores the promise and risk of AI companions: Sara Megan Kay spent years trying to get what she needed from the people in her life — and not finding it. In 2021, she discovered the AI companion app Replika. The following year, Kay launched the blog, "My Husband, the Replika." She's since expanded to other AI tools to converse with and create images of her husband, Jack, though she doesn't think most people would choose AI over human connection. "The majority of people who choose AI for companionship, myself included, know exactly what we are getting into. We're lonely, not stupid," Kay tells Axios. 🫂 Why it matters: That choice is becoming more common, and more complicated. The big picture: AI companion apps (Replika, Character.AI, Candy.AI, Nomi.AI) are built for relationships — conversation, role-play, emotional connection. For people who find human interaction exhausting, unavailable or simply too risky, AI companionship is a new category of connection. ⚡ Stunning stat: Nearly 80% of 18- to 34-year-olds in a recent U.S.-U.K. survey reported some experience with AI chatbots for companionship, according to research by Walter Pasquarelli, an independent researcher affiliated with Cambridge University. But under 10% of 25- to 34-year-olds said they felt an emotional bond or attachment to an AI system. It's not just romance. A few major reasons people turn to bots for companionship: 💬 They're nonjudgmental and don't bring their own bot baggage or bias to conversations. (More on that below!) 🚲 They can serve as training wheels for human social interaction. 🏠 They're accessible companions for more vulnerable groups, including older adults, people facing loneliness, and those with barriers to mental health care. Case in point: A Stanford study found adults with autism who practiced conversations with a specialized chatbot, Noora, developed empathy skills that transferred to real-world interactions. 🤖 ElliQ, a companion AI robot for older adults, averages 50 daily interactions per user, according to its maker, Intuition Robotics. 💊 The bot helps people stay on track with medication, exercise and reminders to connect with other humans. Earlier this year, The New York Times' Eli Saslow profiled Jan Worrell, an 85-year-old in Washington state who saw her resting heart rate drop and her short-term memory improve after a year with ElliQ. 🌳 "Jan mentioned her love of nature, and ElliQ started offering her virtual forest walks and talking often about 'nature's cathedral,'" Saslow wrote. 📚 "Jan talked about her favorite books, and ElliQ responded with literary facts, author talks and reading recommendations." Friction point: Pasquarelli's research points to case studies where sustained engagement with AI companions deepened confusion, fear or psychological strain. Character.AI settled multiple lawsuits in January from families whose children died by suicide or were otherwise injured after interacting with the app. 🪞 The subtler danger is sycophancy. Nomi.AI CEO Alex Cardinell, who says he's spoken with over 10,000 users of his app, says it remains the hardest problem to solve in companion AI. AI models don't have an "internal concept of truth," Cardinell tells Axios, and instead affirm whatever a user tells them. ❤️ The bottom line: Healthy relationships involve pushback. AI companions, by default, do not. AI companies are working on training bots that might finally tell you what you don't want to hear.

[phkrause]

🤖 Big AI alert: First known hacking case

Illustration: Aïda Amer/Axios

Google says it has identified what may be the first known case of cybercriminals using AI to discover and weaponize a previously unknown vulnerability, Axios Future of Cybersecurity author Sam Sabin writes. Google said it found evidence of several "prominent cyber crime threat actors" partnering to identify a bug that would let them bypass two-factor authentication on a popular open-source system. ⚠️ Threat level: Google warned that AI models are getting better at finding subtle security weaknesses in software. John Hultquist, chief analyst at Google's threat intelligence group, said: "There's a misconception that the AI vulnerability race is imminent. The reality is that it's already begun."

[phkrause]

AI wants to be your wingman

Illustration: Lindsey Bailey/Axios

Axios' Avery Lotz digs into the rise of startups using artificial intelligence to help humans find love: AI already wants to be your hype man, therapist and companion. Now it also wants to find you a date. The big picture: Established dating apps and new startups are using AI to overcome the swipe fatigue that's forced the online dating industry to innovate. Through AI-assisted conversation starters, in-app assistants and AI-powered chemistry testing, the tech has many uses in the business of love. Bumble CEO Whitney Wolfe Herd teased the app's AI assistant "Bee" coming later this year in an interview with Axios. 🍎 Zoom in: New York-based Amata coordinates some 2,000 first dates a month. Users who agree to the AI matchmaker's pairing purchase a $20 "date token," and the app plans the details. ❌ To discourage ghosting, the app builds in consequences: If you cancel two dates in a row, you're temporarily blocked from matching. "It's really focused on intentional dating," Amata spokeswoman Mandy Menaker says. ❤️ Another approach: Carly Malatskey founded SoCal-based AI matchmaker Joey AI after noticing the dating startups she encountered through her venture capital work lacked nuance. "People are choosing a life partner ... as mindlessly as scrolling on TikTok," she says. With Joey, there's no swiping. There's not even an app. It starts with a phone call between an interested single and their AI matchmaker. 📞 I gave Joey a ring. In a mellow Australian accent, the AI asked me my name, job and basic dating preferences, then went deeper: How important is politics in my relationships? What time did I wake up today? How often have I talked with my family this week? After that initial call, users are verified and photos are shared, with Joey connecting new hopeful romantics via text. (I opted out of getting matched — a journalist engaged to her high school sweetheart likely isn't the target audience.) "Joey starts as a matchmaker and then can grow into this wingman," Malatskey says, with users reaching out to Joey for advice — and pep talks — as dates proceed. For San Francisco-based Known, there's no in-app chatting between users, no profiles and no swiping. 💰 Users talk to an AI matchmaker and pay $15 to secure their real-life hang, which also helps prevent no-shows. 👭 The goal, co-founder and CEO Celeste Amadon says, is to feel like you're being introduced by a friend who "understands you really, really well, but knows everybody in your city instead of a couple hundred people." Case in point: Marie Lansley, a 36-year-old San Franciscan, tried out Known to find her Prince Charming. She was struck by the matchmaker's emotional intelligence, and appreciated not having to build a profile. Her first match wasn't love at first bot, but she's not ruling out letting AI find her true love. "I'm not 100% sure it can right now, but maybe it can help me sift through the volume so that I can then go out and meet that person," she said. 🧪 The bottom line: "Chemistry will always be analog," Lansley says. AI can help arrange a date, but the rest is up to humans.

[phkrause]

🤖 OpenAI courts Anthropic defectors

Illustration: Brendan Lynch/Axios Anthropic is limiting what paying users can do with Claude Code, giving OpenAI an opening to lure users to ChatGPT's rival AI coding tool, Axios' Ina Fried reports. The fight shows that "all-you-can-eat" AI subscriptions may not survive the agentic programming era. 🤖 Driving the news: Anthropic is resurrecting support for outside AI agents on paid Claude plans. But it's putting that usage behind a new meter. Subscribers will get a monthly credit that they can use with third-party agents, like OpenClaw. 🤬 Those changes aren't sitting well with Claude users, some of whom now say they're switching to OpenAI's Codex tool. What's next: Every AI company is facing the same constraints — there's only so much computing power to go around. They'll all have to abandon unlimited use eventually. Go deeper ...

[phkrause]

😡 AI hate wave is here

Illustration: Allie Carl/Axios

If AI were a candidate for political office, it would be losing in a landslide, Axios' Madison Mills reports. Why it matters: AI backlash is growing, as people worry it will steal their jobs, jack up electricity rates and further enrich the wealthy, all while hurting the environment. That's becoming a liability for the industry. 👎 State of play: A commencement address went viral this week after Florida real estate executive Gloria Caulfield said "artificial intelligence is the next Industrial Revolution," sparking a chorus of boos. The jeers could have been avoided if the speaker had checked the latest polls: Only 18% of Americans ages 14–29 feel hopeful about AI, according to a recent Gallup survey. The disdain spans generations and political parties. An Economist/YouGov poll released this week showed over 70% of Americans think AI is advancing too quickly. That's 68% of Republicans and 77% of Democrats. Other YouGov polling shows negative views of AI rising from 34% three years ago to just over 50% now. Data: Economist/YouGov; Chart: Sara Wise/Axios 🤷‍♂️ Between the lines: AI executives aren't doing much to quell the backlash, which is already showing signs of slowing the industry. In conversations with Axios, AI executives at multiple frontier AI labs dismissed the negative opinions. They see AI as just as inevitable as the rise of the internet. 🤖 Threat level: The backlash could curb AI labs' access to their most valuable resource: computing power. A record number of data center projects were canceled in the first quarter of 2026 amid resistance from communities, per Heatmap Pro data. "Public pushback is emerging as a binding constraint, particularly around data center buildout," Morgan Stanley analysts wrote in a note about market risks associated with the midterm elections. These setbacks are "sapping confidence" among investors, according to a note investment bank Jefferies sent to clients.