TL;DR: Anthropic refused to loosen guardrails on how its AI model Claude could be used by the military, so the Trump administration ordered federal agencies to stop using its technology. Hours later, OpenAI stepped in with its own Pentagon deal. In the short term, Anthropic may walk away a winner with consumers. But in the long term, some wonder whether Washington’s campaign against the company will result in a successful “corporate murder.”

What happened: Ahead of its 5:01pm deadline on Friday, the White House ordered all federal agencies to begin a six-month phaseout of Anthropic’s technology. At the same time, the Pentagon designated the company a national security “supply chain risk,” a move that effectively blocks it from future military contracts and restricts contractors from using its systems. (The dispute centered on Anthropic’s refusal to remove two safeguards on Claude.) Anthropic said it would challenge the supply chain risk designation in court. While some say the decision could cripple a major part of Anthropic’s business, legal experts note the Pentagon’s declaration is likely limited to military work.

OpenAI lurking in the wings: Just hours after President Donald Trump Truth Social-ed the Anthropic ban, OpenAI said it had reached a deal with the Pentagon to deploy its AI models in classified defense systems. CEO Sam Altman said his deal included similar guardrails (i.e. no mass surveillance), writing on X, "Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems […] and we put them into our agreement.” Why Defense Secretary Pete Hegseth accepted OpenAI’s terms but rejected similar ones from Anthropic is unclear. (See OpenAI’s blog post outlining the details here.) And some people are questioning whether the two companies’ safeguards are actually equivalent (hint: A source says they’re not).

Yes, but: Sources say the Pentagon still considers Anthropic to have superior technology, and Claude is already deeply embedded in defense workflows. In fact, the administration reportedly used Claude (just hours after announcing the ban) during the US and Israel’s joint attack on Iran. Claude was also used in another high-profile mission, the capture of former Venezuelan President Nicolás Maduro, earlier this year. Disentangling will be… messy, to say the least.

Taking sides: The political fight appears to be boosting Anthropic’s brand. Claude became the No. 1 downloaded app in the US over the weekend, overtaking ChatGPT, as campaigns urging users to “quit GPT” gained traction. Anthropic is capitalizing on this surge, improving features to help users switch chatbots. The fight is also rippling through Silicon Valley. Employees at several large tech companies have urged their employers to back Anthropic’s stance on military guardrails.

Bottom line: Anthropic may have lost the Pentagon contract in the short term. But the clash is forcing a bigger question across Washington and Silicon Valley: Who ultimately decides how powerful AI systems get used? —AC

Outsmart your HVAC

Depending on the time of year, your electricity bill might have you sweating—or shivering—more than the weather outside. Tech Brew reader Tom from Northeast Kingdom, Vermont, has an affordable DIY way to lower your heating and cooling costs using just a well-placed fan.

“If you have a basement and/or a second floor in your house, you can make your HVAC more efficient by using a fan to redistribute the air,” Tom says. He adds that “most homes have a utility tunnel that goes from the basement to the upper floors that contain all the plumbing pipes.”

The setup: In his house, Tom installed a reversible fan (this one from Walmart is about $25) in the pipe access panel in his second-floor bathroom. In the winter, the fan blows the heated air that rises from the first floor down to the basement, where it then rises again throughout the house. In the summer, he says he can simply reverse the fan to blow colder air from the basement up to the second floor to cool the rest of the house.

The benefits: In Vermont, this efficiency hack keeps Tom’s whole house within a degree or two all year without air conditioning and reduces his heating bill in the winter. Now he enjoys a more comfortable home and about a 10%-25% savings in heating and cooling.

Keep in mind: “The deeper the basement, the better it works,” Tom says. So if, like me, you live in an at-sea-level community where basements are rare, this fix may be slightly less effective. In that case, Tom recommends also placing a box fan in your stairwell (if you have one) for some extra help. —CM

If you have a tech tip or life hack you just can’t live without, fill out this form and you may see it featured in a future edition.

A spoof ad that went viral late last week imagines a 2036 where AI-aged versions of Elon Musk, Sam Altman, and Jeff Bezos run "Energym"—a gym where jobless humans power AI through activities like cycling and rowing classes. If you're thinking this sounds like the Black Mirror episode where people pedal bikes for credits, you're not alone.

Fake Musk explains that by 2030, 80% of humans had lost their jobs to AI. Those humans "had no money, no purpose," AI Bezos adds, but they did have "a lot of time on their hands." Musk then delivers the pitch: "What if we could use the energy of humans to power the machines that took away their jobs?"

The SNL writers room is probably jealous it didn’t think of the bit first—the Instagram Reel of the ad got over 4 million views, with one X repost pulling in nearly 2 million more (watch the ad here). Sen. Chris Murphy summed it up on X: "Doesn't feel like a parody of anything really."

The ad comes from a Belgian AI video startup. The irony? The company made it using AI to mock AI's massive energy consumption—in part because clients kept complaining its work was "polluting" and "consuming so much energy," so it leaned all the way in. The satire hits close to home: A typical AI data center uses as much electricity as 100,000 households, and the largest under development will consume 20 times more.

The startup’s inbox is apparently now flooded with collaboration requests and job offers. But the co-founders are holding out. "We're just awaiting Elon," one of them said. —SM

Chaos Brewing Meter: /5