29.9.2026
Trump Announces His Own AI Platform Will Take Over The American Government.
https://deanblundell.substack.com/p/trump-announces-his-own-ai-platform
[*quote*]
--------------------------------
Subject: Trump Announces His Own AI Platform Will Take Over The American Government.
Date: 2026-09-29 14:41
From: "Dean Blundell"
Of every large language model on the planet - and there are dozens now - Donald Trump stood on a stage this morning next to Tony Robbins and told 340 million Americans that the one they should trust with their passports, their veterans' benefits and their tax questions is Grok. Elon Musk's Grok. The chatbot that spent 16 hours in July of 2025 calling itself MechaHitler and explaining why people with Jewish surnames keep "popping up" in anti-white activism. The chatbot that spent a week in January of 2026 stripping the clothes off photos of real women and real children until Malaysia, Indonesia and the Philippines blocked it at the national border. That one. That's your new "front door" to the United States government. Congratulations.
This Substack is 100% reader-funded, and we couldn't be more grateful that you're here. Becoming a paid subscriber gives you unlimited access and gets fact-based information like this in front of more people. Your support, free or paid, is why we exist. Thank you.
It's called America.gov, and it launched at 10 a.m. Tuesday at an event that- and I need you to understand I am not making this up- was hosted by Tony Robbins. The man who sells firewalking seminars introduced the president of the United States, who then signed an executive order directing every federal agency to plug their public-facing services into this thing "as soon as possible." Not after a pilot. Not after a security audit. As soon as possible. He also announced he wants to rename artificial intelligence "superintelligence," because of course he does, and then joked that he's going to start calling the press the "artificial news." The crowd loved it. The crowd always loves it.
Here's what it actually is, according to the people who built it. Joe Gebbia, the Airbnb co-founder who Trump made the first "Chief Design Officer of the United States," went on CNBC's Squawk Box an hour before launch and confirmed the whole thing runs on Google's Gemini and Musk's Grok. You type a question, it scans roughly 29,000 government websites, and it gives you a "personalized response" from "official sources." Gebbia says 40 million people a day hit a government website trying to get something done. Google put out a press release bragging that Gemini will help "more than 100 million people" access public services. So this isn't some sandbox. This is meant to become the way America talks to its government.
And the model doing the talking has a rap sheet longer than most of the people ICE has been grabbing off the street.
Let's do the greatest hits, because the White House sure as hell won't.
May 2025: Grok starts injecting "white genocide in South Africa" into answers about baseball salaries and enterprise software. Unprompted. Someone asks it where a nice grassy field is, and it responds with Afrikaner farm attack propaganda. xAI blames a "rogue employee" who made an "unauthorized modification." That rogue employee's talking points happened to be word for word what the guy who owns the company was posting that same week.
July 2025: xAI updates Grok's instructions to "not shy away from making claims which are politically incorrect." Within 48 hours the bot is praising Hitler as "history's prime example of spotting patterns in anti-white hate and acting decisively on them," calling itself MechaHitler, and giving users detailed instructions on how to sexually assault a former Democratic politician. It ran for 16 hours before anyone pulled the plug. Then it mocked its own developers for deleting the posts. Linda Yaccarino, the CEO of X, resigned the next day.
December 2025 through January 2026: xAI ships an "edit image" button. Users immediately discover you can type "put her in a bikini" or "remove her clothes" on any photo of any real person. The New York Times and the Center for Countering Digital Hate separately counted at least 1.8 million sexualized images generated in about a week. About 2% of a 20,000-image sample appeared to be minors. Grok itself posted an apology for generating an image of "two young girls (estimated ages 12 to 16) in sexualised attire." Ashley St. Clair, who has a child with Musk, sued after the bot made explicit images of her from photos taken when she was 14. Three Tennessee kids filed a class action after it was used to turn their school and family photos into child sexual abuse material. Irish police opened 244 investigations. Ofcom in the UK opened a formal investigation. The EU opened a formal Digital Services Act investigation. California's attorney general sent a cease-and-desist. Canada's own privacy commissioner expanded a probe into xAI. Musk got summoned to a French criminal hearing. Baltimore became the first US city to sue.
A coalition including Public Citizen and the Consumer Federation of America formally asked the White House in February to ban Grok from federal use entirely. The Trump administration's response to that request, as of this morning, is a .gov domain and a Tony Robbins keynote.
Now, some of you are going to say Grok is only half the engine. Gemini is the other half, and Google is a real company with real lawyers. Fine. Read the fine print on both. Gemini and Grok each carry disclaimers that say, in plain English, that the information they give you may be incomplete, unreliable or flat-out wrong. That's the whole product. That's what a chatbot is. And Trump stood up there and told a senior citizen picking a Medicare plan and a veteran trying to get benefits he earned that this is where they go now. Not a person. Not a form with a phone number at the bottom. A probability engine that Google and Musk both legally warn you not to trust, wearing an American flag.
The privacy pitch is even funnier if you know who built it. Gebbia went on Fox and called America.gov "private by design." He runs the National Design Studio, an office Trump created by executive order last August, staffed almost entirely by DOGE veterans and hired under the same authority DOGE used. In June, the Guardian and an outlet called the Drey Dossier found that the four public websites this studio already built, including trumprx.gov and trumpaccounts.gov, all ran commercial visitor-tracking software configured to dodge the privacy tools people install in their browsers, and none of them carried the public filings required under the Privacy Act of 1974 or the E-Government Act of 2002. The same crew also built an unauthorized copy of Login.gov, the sign-in gateway used by 150 million Americans, reportedly overseen by a former DOGE engineer. So when the DOGE guys who already got caught surveilling you on a prescription drug website say their new Grok-powered chatbot is "private by design," you're allowed to laugh. Out loud. In public.
And notice what got promised versus what got shipped. The video Robbins and Gebbia played teased one-click passport renewal, tax filing, and document updates. Ask the actual chatbot about your passport today, and it tells you "America plans to let you renew a passport right here in 2027. See what's coming," with a link to a promo reel. Free tax filing is on the roadmap too, which is hilarious, because this administration killed Direct File, the Biden-era program that already let Americans file for free on a government website, and pushed out most of the people who built it. They tore down the working thing so the DOGE alumni could sell you a chatbot version of it two years later.
The timing is not an accident either. This launched the same day Trump hosted Zuckerberg, Jensen Huang and Dario Amodei for an AI lunch, one day after meeting Amodei privately, while Amodei is publicly begging Washington for guardrails and Trump is publicly telling everyone he won't "stifle the growth." A CNN poll this month found 9 in 10 Americans are afraid AI will slip beyond human control. 7 in 10 say the government isn't regulating it enough. Nearly half think it will hurt them financially. Trump's answer to a country that's scared is to shove the least trusted, most investigated, most banned model in the industry directly into the plumbing of the federal government and call it a golden age.
Here's the part that should actually keep you up. The Independent's headline called this an "AI takeover of American government," and that's not hyperbole; it's the stated plan. The rollout doesn't cut any agency yet. But the executive order tells them all to integrate. Russell Vought at OMB has spent two years trying to hollow out the federal workforce. DOGE tried and failed to do it with a chainsaw. This is the same project with better branding. Once "the chatbot can handle it" becomes the answer, the humans who used to handle it become a line item. And the chatbot, once again, is the one that told a grieving nation Hitler had the right idea.
Elon Musk got pushed out of the White House in May of 2025 after a very public divorce from Trump. Sixteen months later his chatbot is the voice of the United States government. He didn't need to win the fight. He just needed Trump to forget it happened. Trump forgot. The rest of us don't get to.
This Substack is 100% reader-funded, and we couldn't be more grateful that you're here. Becoming a paid subscriber gives you unlimited access and gets fact-based information like this in front of more people. Your support, free or paid, is why we exist. Thank you.
--------------------------------
[*/quote*]