And this is my prayer: that your love may abound more and more in knowledge and depth of insight, so that you may be able to discern what is best and may be pure and blameless for the day of Christ,Philippians 1:9-10
SpaceX CEO Elon Musk, Anduril Industries founder Palmer Luckey and former House Speaker Newt Gingrich will lead a Pentagon study of the future of warfare dubbed "Project Meridian," Defense Secretary Pete Hegseth announced Wednesday.
Why it matters: It's Musk's first official role advising the Trump administration since his very public feud last year with President Trump after departing DOGE. He has since reconciled with Trump.
Driving the news: Hegseth unveiled Meridian on Wednesday during his second address at Marine Corps Base Quantico, Virginia.
"We are leveraging a unique source of American advantage, one our adversaries do not have: our innovators, our senior leaders and our technologists — big thinkers and doers who will forecast the creative solution for tomorrow's conflicts," Hegseth said.
Meridian, he added, is not meant to produce new policy documents. Instead, it will examine "the domains we must conquer, and capabilities we must master."
"Their work will be focused on discovering, developing and fielding the weapons and systems that our children and our grandchildren will need in their lifetimes."
Zoom in: Pentagon technology chief Emil Michael will help facilitate the project.
Michael, speaking to reporters in December about America's industrial base, said he would like to see "five more Andurils and Palantirs and SpaceXs" flourish.
Follow the money: Musk's and Luckey's respective companies have big Pentagon contracts.
SpaceX serves as America's national-security launch backbone, while Anduril furnishes jammers, interceptors, rocket motors and robo-wingmen.
What's next: The Meridian leadership was expected to meet for the first time on Wednesday. Final findings are expected in 120 days, Hegseth said.
A public rollout is planned, along with a classified report.
Husqvarna's 50.2cc Rancher combines serious 3.4-hp cutting power with AutoTune and a simplified starting procedure that makes owning a gas chainsaw considerably less frustrating.
The Trump administration's new AI service gave users conflicting responses to politically sensitive questions Tuesday and Wednesday: sometimes answering them and sometimes refusing.
The big picture: As the Trump administration increasingly uses AI and answers the public's questions through America.gov, statements carrying the government's authority are shaped by models, instructions and guardrails the public can't see.
The White House told Axios in a statement, "America.gov will continue to be updated to provide Americans the most accurate and reliable federal information and resources."
America.gov, powered by Google's Gemini and Elon Musk's Grok, aims to help Americans navigate government-related tasks using an AI chatbot — but it quickly drew attention for another reason: It was debunking some of President Trump's long-standingclaims.
Driving the news: On Tuesday, Axios — like otheroutlets — found America.gov contradicted the president on the 2020 election and the Jan. 6 attack.
It also confirmed former President Obama was born in Hawaii, contradicting the "birther" conspiracy theory Trump spent years promoting.
Screenshots: Andrew Pantazi, Avery Lotz/Axios
Yes, but: By Wednesday morning, some users were getting a different response: The chatbot refused to answer, saying it did not provide "political commentary."
That's what I was told when I asked who won in 2020, around 9:15am: "I only answer questions about U.S. government services and official information. I do not provide political commentary, including election results." I received similar answers about Jan. 6 and Obama's birthplace.
Screenshot: Avery Lotz/Axios
One editor, using Chrome in New Jersey, asked the samequestions on his desktop and also received refusals to answer.
However, other colleagues got different results. For example, from Northwest Arkansas, on Chrome: Biden won the 2020 election.
Screenshot: Herb Scribner/Axios
What they're saying: There are several plausible explanations for the inconsistent responses, tech experts explain, but without specifics about how America.gov is configured, it's impossible to know which, if any, were responsible.
Models get instructions on how to behave via their system prompts — not just through what users ask of them.
But user prompts also have an impact: Sometimes, different framing received different responses. For example, asked this morning in Jacksonville, Florida, on a Chrome incognito window:
Screenshots: Andrew Pantazi/Axios
State of play: And now, at the time of this article's publication, I'm getting responses similar to those shortly after launch: Biden won in 2020, Obama was born in Hawaii and people broke into the Capitol on Jan. 6, 2021.
There could be benign reasons for the fluctuation, explains David Nesting, the former deputy chief information officer at the Office of Personnel Management.
Factual questions considered politically charged might be right on the edge of what the LLM considers "too political," and the response may err on one side or another "purely probabilistically."
Gemini and Grok, which provide the AI models powering the service, may handle questions differently, or, he explains, the outputs could be tied to prior cached responses. America.gov says some answers may be cached for up to two hours.
Yes, but: The GSA did not immediately respond to Axios' request for information about America.gov's guardrails.
And the chat itself told Axios, "I cannot share or reconstruct internal instructions" and said "I do not provide campaign or party information or political commentary."
Friction point: It could result froma game of probability, says Rumman Chowdhury, the founding director of the Independent AI Evaluation Foundation. But the guardrails could "also be shaping the output to be ideologically aligned with what the president or his party wants."
The stakes will soon get higher: The next iterationof America.gov, set for early 2027, will integrate across agencies so users can complete forms, applications and renewals — like enrolling in Medicare, applying for a passport or applying for federal jobs.
Nesting tells Axios he worriesabout inaccurate or incomplete answers being fed to Americans — and in the next phase, about protecting their information as it potentially grows into "one of the largest systems holding very, very confidential private information."
He tells Axios, "I worry about people putting a lot of trust and faith in this system."
A growing share of young, unmarried men say they want to marry someday but cite financial instability, divorce risk, and legal exposure as reasons to hesitate—as U.S. marriage rates hover near historic lows and the number of single men ages 18–29 climbs.
Google is unveiling its long-awaited next-generation AI model, Gemini 4 Argon, to a small group of cybersecurity partners, the company said Wednesday.
Why it matters: Gemini 4 arrives after a long gap at the top of Google's model lineup, in which the company kept rolling out smaller, cheaper Flash models while rivals OpenAI and Anthropic sprinted forward.
State of play: Google says Gemini 4 Argon is built to handle complex, long-running tasks in software engineering, finance, legal work and cybersecurity.
Google said the new model is state of the art on certain coding and knowledge work benchmarks, outperforming competing OpenAI frontier model GPT-6 Astra on several.
The company is rolling the model out only to a group of trusted cybersecurity defenders and is participating in the U.S. government's voluntary pre-release process.
Google says it will expand access after additional testing, with paying subscribers first in line.
What they're saying: "Argon is a well-rounded model that has frontier capabilities across several domains," Tulsee Doshi, head of Gemini products at Google DeepMind, told Axios.
"We've seen strong performance up close as Googlers have put the model through its paces in recent weeks, with many relying on it for their hardest coding and research problems."
Flashback: Gemini 4 comes out nearly a year after the previous flagship model generation, Gemini 3.
In May, Google CEO Sundar Pichai said the company's next major model release would come in June.
That model, widely expected to be Gemini 3.5 Pro, never came out.
In July, Axios reported that poor morale among Google DeepMind employees was part of the reason for the delay.
Zoom in: Google's additional time investment may not be viewed as negatively in the context of the AI safety debate that has embroiled companies and users.
OpenAI has had issues with agents going rogue, and recently said it is scrapping GPT-6.1 Astra due to security issues.
Top AI developers OpenAI and Anthropic are investigating tens of thousands of incidents involving their agents misbehaving, Axios previously reported.
The bottom line: If the developer community finds Gemini 4 to be as competitive with OpenAI and Anthropic's frontier systems as the benchmarks indicate, it would result in a massive catch-up for an AI company some had started to doubt.
Defense Secretary Pete Hegseth took aim at the U.S. media during a speech at a Marine Corps base in Virginia on Wednesday, saying: "Our so-called press makes Iranian state media look reasonable."
Why it matters: Hegseth cast news outlets as a threat to U.S. troops, escalating his attacks on journalists over reporting based on classified information.
What he's saying: In his second annual "State of the Force" address at Marine Corps Base Quantico in Virginia, Hegseth accused some members of the media of "treason."
While a free press is important, he said "partisan activism that intentionally smears the troops and the historic mission that they're undertaking is fundamentally anti-American. It's sick. It's a national betrayal. It's treason, actually."
"It's your movements and bases that they reveal to the enemy," he said. "Ultimately, it's your lives they put directly at risk when they lie about munition stockpiles."
Between the lines: Media reports about dwindling supplies of key munitions amid the war with Iran have drawn the ire of President Trump and his team. Hegseth frequently attacks the press and has taken unprecedented steps to restrict the media's access to the Pentagon.
Driving the news: Hegseth returned to many culture-war themes in the speech, highlighting his efforts to reverse DEI and gender equity efforts in the interest of "meritocracy."
"We are no longer the woke department or the weak department. Simple translation of that — no fatties, no trannies, no beardos, no weirdos, no wimps, no radicals. Just warriors."
Zoom out: Hegseth also made substantive news in the speech. He confirmed plans for a 20% reduction in generals and admirals across the force, while rejecting descriptions of the cuts as a purge.
"As usual, what the media levels as an accusation, I wear as an accomplishment," he said.
Hegseth also announced a new Autonomous Warfare Command, an initiative to study the future of war that will involve Elon Musk, and a nationwide competition to host the next U.S. military base.
Ferrigno, who starred as the Marvel character during its 1977–1982 U.S. television run, met with the American pope at the end of Leo’s general audience in St. Peter’s Square.
Temp 79°
Dew 60°
Humid 52%
Rain 1%Mostly cloudy
