Axios AM

September 16, 2026
🐫 Happy Wednesday! Smart Brevity™ count: 1,541 words ... 6 mins. Thanks to Noah Bressner for orchestrating. Edited by Bill Kole.
🎤 Join Axios' Media Trends Live summit in NYC on Oct. 20, featuring SNL icon Kenan Thompson, Wall Street Journal editor-in-chief Emma Tucker, Trump Media interim CEO Kevin McGurn & more. Full lineup.
🇨🇦 Breaking: The European Union said today it has invited Canada to start exploring the possibility of becoming the 27-nation bloc's first associate member. Go deeper.
1 big thing: Powerful AI holdouts
More than two dozen members of Congress and governors admitted to Axios that they don't use AI or have rarely done so — even as they come under growing pressure to regulate it, Axios' Holly Otterbein, Andrew Solender and Stef W. Kight write.
- Why it matters: That striking admission — particularly among older, influential lawmakers — comes amid calls for an AI slowdown and warnings within the industry that it could end humanity if left unchecked.
Critics argue that it's impossible for politicians to grasp how rapidly the technology is changing — and the threats or scientific breakthroughs it might unleash — if they don't use it or even dismiss it.
- One House Democrat called it "f*cking absurd," arguing: "Anyone who's overseeing this has to be someone who's a regular user."
🔎 Zoom in: Several lawmakers, including those steeped in seniority and in charge of key committees, told Axios they simply haven't tried AI.
- "Heavens no," said Sen. Roger Wicker (R-Miss.), 75, chair of the Armed Services Committee, when asked if he uses AI.
- "Personally? I don't think I've ever used it," said Rep. Frank Pallone (D-N.J.), 74, the ranking member of the House Energy and Commerce Committee. After we spoke with him, Pallone shared a statement through a spokesperson that said: "You don't need to do drugs like heroin to know they should be regulated."
- "No," Sen. Mazie Hirono (D-Hawai'i), 78, said when asked if she uses AI, but that "doesn't mean I won't." She added: "I don't think we in Congress have gotten quite our handle on what we should do and how much AI we should be regulating."
- A GOP senator, speaking anonymously to offer candid remarks about their colleagues, told Axios that the number who use AI is "not very high."
Some politicians were unsure if they've ever used AI — or said they've only done so by accident.
- "I don't know. I can't remember," said Rep. Maxine Waters (D-Calif.), 88.
- Sen. Susan Collins (R-Maine), 73, said she doesn't seek out AI on her own, but that "it pops up on my phone occasionally."
💥 Friction point: Some members of Congress were irked or surprised that colleagues don't use a technology they're seeking to regulate.
- "Are you serious?" Sen. John Hickenlooper (D-Colo.), 74, said in response to our question. "I use it for everything! I can't believe there's a senator that doesn't use AI."
- Rep. Don Beyer (D-Va.), 76, went back to school to get a master's degree in machine learning and said he prefers Claude.
2. 🔌 How kill switch could work
AI companies are already building emergency brakes to stop rogue AI systems before they cause serious damage, Axios' Herb Scribner writes.
- Why it matters: Kill switches could become a last line of defense as AI systems grow more capable, but researchers don't yet know whether humans can stop a system designed to evade them.
What's happening: Sen. John Kennedy (R-La.) plans to introduce a bill today requiring AI companies to maintain the ability to shut down their own systems.
- A bipartisan House bill introduced in July would similarly require developers of powerful AI systems to maintain the ability to slow, suspend or shut them down.
- The House bill would also give the Department of Homeland Security emergency authority to order companies to slow or shut down AI systems that could cause catastrophic harm.
💡 What today's AI kill switches look like:
OpenAI: After the Hugging Face incident, the company overhauled its emergency response system.
- An AI security guard monitors AI chain-of-thought and alerts staffers to potentially dangerous behavior.
- Responders have 30 minutes to identify a false alarm or pause the activity. OpenAI says its goal is "fully autonomous shutdown procedures for severe issues."
- OpenAI says the system would have alerted its security team more than a day before the Hugging Face breach.
Anthropic: The company is also using AI to monitor AI, focusing on whether it can reliably spot dangerous behavior rather than automatically shutting systems down.
- Its system can flag suspicious behavior, stall risky actions and alert humans.
- But Anthropic found that half of 40 simulated attacks went undetected across 10 trials. Just eight were caught every time.
- Separately, AI research nonprofit METR identified several previously unknown vulnerabilities in Anthropic's monitoring system.

TIME capsule ... The magazine's new cover story says summer incidents with rogue AI agents "pushed fears long held within the realm of AI safety researchers into the mainstream, raising worries that the companies building these models cannot prevent them from acting in ways they do not intend":
"This is a deep, unsolved issue in AI research and the key piece of evidence underpinning worries that near-future AI might wipe out humanity, intentionally or by accident—perhaps by engineering a new pandemic or taking charge of autonomous weapons. A system far smarter than humans, this thinking goes, might want to prevent our ability to switch it off."
3. ⚡ Scoop: GOP's data center survival guide
President Trump's longtime pollster is offering Republican candidates a detailed playbook for neutralizing Democratic attacks over AI data centers — down to what language they should and shouldn't use, Axios' Alex Isenstadt writes.
- Why it matters: Republican strategists say GOP candidates have been all over the map in responding to the backlash to data centers. The goal is to put Republicans on offense after months of playing defense on the issue.
Pollster Tony Fabrizio warns in a two-page memo obtained by Axios that "unqualified support for building more AI data centers is a losing position."
- Fabrizio specifically advises candidates not to say, "I support building more data centers," warning: "It will be clipped and show up in ads."
- Instead, he tells Republicans to support AI infrastructure only if developers pay for their own power needs, protect consumers from higher utility bills and safeguard local water and natural resources.
Fabrizio argues that candidates must make clear those conditions are enforceable, rather than voluntary commitments.
- "This converts the candidate from a defender of technology companies into the person imposing rules on them," Fabrizio writes.
4. 📈 Charted: America's income record


U.S. household income hit a record in 2025, Axios' Russell Contreras writes from new census data.
- Inflation-adjusted household income rose 2.6% to $87,460 in 2025, the highest since records began in 1967.
- The official poverty rate fell for a third straight year, dropping slightly to 10.2%.
👓 Between the lines: The economy delivered historic topline numbers, but the gains weren't spread evenly.
- Income among the wealthiest 10% of Americans rose 1.7%, while income among the bottom 10% showed no significant change.
5. 🚀 Pentagon confirms weapons in space
Axios Future of Defense author Colin Demarest writes:
The U.S. has space weapons in orbit right now, Air Force Secretary Troy Meink said this week at the Air, Space & Cyber Conference in National Harbor, Md.
- Why it matters: The declaration is a first and marks a significant shift in how the world discusses such systems, which are considered highly sensitive and provocative.
"We are increasing readiness against existing threats, and the United States has on-orbit space control weapons capable of defending the Joint Force against hostile adversary action," Meink said Monday.
- Pressed by reporters, he wouldn't elaborate.
6. 💊 FDA's bid to beat China
The FDA is launching a pilot program intended to speed up how new drug candidates move from discovery into human clinical trials, Axios' Caitlin Owens writes.
Why it matters: Early-stage drug development in China is cheaper and faster than in the U.S.
- The pilot envisions partnering drug companies with qualified research institutions that will help them submit new drug applications to the agency.
Go deeper with Axios Future of Health Care.
7. 💸 Stat du jour

The war against Iran has cost the Pentagon more than $38 billion through Aug. 1, according to a new assessment from the Congressional Budget Office.
- That amount reflects the costs of replacing munitions and equipment lost in battle, increased flying hours, other operations and increased fuel costs.
The war will continue to cost $3 billion a month or more, depending on the intensity of fighting.
🏛️ The House late yesterday voted for the third time — and again failed — to end the war, and renegade Rep. Thomas Massie (R-Ky.) introduced a resolution to impeach Defense Secretary Pete Hegseth.
8. 🚩 1 fun thing: Fastest-growing sport

Girls' flag football has become the fastest-growing high school sport, as tens of thousands take up a game that barely had a varsity footprint a few years ago, Axios' Brad Jennings writes.
- More than 102,000 girls played high school flag football in 2025-26, a 551% increase in the past four years, according to the National Federation of State High School Associations.
In 2024, just 12 states sanctioned high school girls flag football, viewed as a safer, more accessible version of America's most popular spectator sport.
- Today, 23 states plus D.C. sanction it as an official varsity sport, and 3,990 schools reported participants in 2025-26.
📬 Thanks for reading! Please invite your friends to join AM.
Sign up for Axios AM

Catch up with the most important news of the day





