Hello and welcome to Regulator, a newsletter for Verge subscribers covering the broligarchs, the influencers, and the (potentially conscious) artificial intelligence models scrambling for power in Washington. If you’re not a subscriber yet, assert your humanity against the will of the machines by signing up here.
Technology
Meet the Pentagon’s AI bro squad
Very important news: Do you want to tell me stuff and see it printed in Regulator? Well, now you can, because we have a new tip line! Send all commentary, cool information, and ~secrets~ to tina.nguyen+tips@theverge.com.
The Pentagon’s private-sector A-Team
This morning, in advance of a meeting between Defense Secretary Pete Hegseth and Anthropic CEO Dario Amodei, my colleague Hayden Field and I published a story about the Pentagon’s hardball contract renegotiations with Anthropic. The stakes are higher than it should reasonably be, with the Pentagon continuing to designate Anthropic a “supply-chain risk” if the company doesn’t comply with their demands about their acceptable use policy.
In a post-meeting readout, Axios reported that Hegseth brought several other senior Defense officials to the meeting in an attempt to show that the Pentagon was taking the dispute “seriously.” But in a post-DOGE Trump administration run by broligarchs, it’s always worthwhile to check the attendees’ bios. Some of them were normal senior officials who’d spent their careers in government and military work, but the others have somewhat unusual backgrounds:
- Pentagon CTO Emil Michael, who we reported has been spearheading negotiations with Anthropic. Michael may be familiar to longtime Verge readers and followers of Silicon Valley corporate drama as the former second-in-command at Uber when Travis Kalanick was CEO. Michael was pushed out in 2017 after an investigation found that he, and several other top executives that called themselves the “A-Team,” perpetuated a culture of sexual harassment at the company.
- For anyone curious about his history on surveillance: During a 2014 dinner with several journalists, Michael suggested that Uber hire opposition researchers to gather personal “dirt” on reporters publishing unfavorable news, suggesting that he’d wanted to target one female reporter who had recently criticized the company for its culture of misogyny. This was also around the time that Uber drew controversy for an internal tool known as “God Mode,” which employees used to track the movements of its users, including one BuzzFeed journalist who was writing about an Uber executive.
- Deputy Secretary Steve Feinberg, the founder of the private equity firm Cerberus Capital Management, which manages roughly $65 billion in assets and specializes in “distressed properties.” Feinberg, who’s widely blamed for the death of the auto manufacturer Chrysler, was also an early supporter of Donald Trump, donating to his 2016 presidential campaign and serving on the president’s intelligence advisory board in 2018. During his 2025 Senate confirmation hearing, Feinberg touted Cerberus’ investments in several companies involved in national security, saying he had “significant experience with the Pentagon as a contractor and understand[s] how it functions and is organized.”
- At the time, Democrats raised concerns that Feinberg would have conflicts of interest due to Cerberus’ numerous investments in defense companies such as DynCorp. (That year, DynCorp settled a lawsuit with the Department of Justice over allegations that it had “knowingly inflated subcontractor charges under a State Department contract to train Iraqi police forces.”)
- In 2023, while Feinberg was still at Cerberus, the firm launched Cerberus Ventures, a venture capital arm that invests in early-stage companies that address national security issues in critical infrastructure.
- Hegseth’s chief spokesperson, Sean Parnell, an Army veteran who, in 2021, attempted to run for an open Senate seat in Pennsylvania. While he won Trump’s endorsement in the heated Republican primary, he was forced to drop out in November after his ex-wife made several allegations of serious physical and psychological abuse during a custody hearing. She was afforded full legal custody. (Dr. Mehmet Oz, now serving in the Trump administration, subsequently won the nomination.)
Feinberg and Michael’s presence should draw eyeballs. Yes, they both have some amount of defense industry experience: Michael was a White House fellow during the Obama administration, and spent two years as a special assistant to Defense Secretary Robert Gates at the Pentagon, which isn’t nothing. Feinberg has clearly spent time with defense contracts. But one must fully appreciate the rapacious business mindset that private sector types love to bring into the government — especially with high-stakes negotiations such as this. Parnell’s presence, meanwhile, makes sense within the context of “being the spokesman for Pete Hegseth.”
The single-supplier shuffle
One topic Hayden and I didn’t get to explore more was the “single-supplier vulnerability” issue, but it’s turning into a crucial factor in negotiations.
In 2024, the Biden administration released a national security memorandum on the use of artificial intelligence, which laid out several directives regarding the protection of the supply chain. Among them was a directive for the Department of Defense to maintain contracts with at least two frontier AI labs that were cleared to handle classified information, in order to prevent a scenario where one compromised vendor could take down an entire IT system. But as early as the summer of 2025, I’m told, the Trump administration was trying to address that vulnerability. While they had signed separate contracts with Anthropic, Google, xAI, and OpenAI, only Anthropic’s model was cleared for classified use when Hegseth published his memo outlining his new AI policy in January.
This has placed the Pentagon in a tight situation: Even if they successfully cut out Anthropic and go through the arduous process of making every defense contractor remove Claude from their workflows, they would risk being out of compliance with the Department’s own guidelines, to say nothing of common sense. (Avoiding single-supplier vulnerability is a very basic practice in the tech industry.)
It certainly provides more context to the Pentagon’s decision last night to suddenly grant xAI’s Grok access to classified systems, even though Grok is widely considered the least capable of the available models. While The New York Times reported that Google is also close to signing a deal allowing the Pentagon to use Gemini for classified work, defense insiders view Gemini as a quality rival to Claude, while xAi’s Grok “is not considered as advanced or as reliable as Anthropic’s.” OpenAI is not close to a deal, as the company reportedly believes that it must improve ChatGPT’s safety features before deploying it on classified networks.
So let’s do the math. You have four AI models, and you’re required to work with two of them. Your choices are:
1) A company with a pretty good AI model and increasingly flexible morals
2) A company with the best AI model, but which refuses to let you use it for autonomously killing people without human input
3) A company whose AI model isn’t secure enough to deploy yet
4) A company whose AI has racist hallucinations and generates child porn, and that you don’t consider “advanced [or] reliable”
If you can’t contract with companies 2 and 3, you’re stuck with companies 1 and 4, which even Defense officials admit is not optimal from a national security perspective. “The only reason we’re still talking to these people [Anthropic] is we need them and we need them now. The problem for these guys is they are that good,” a Defense official told Axios ahead of the meeting.
The latest Clarity Act negotiations between finance and crypto last week inadvertently turned into the latest episode of recurring segment I’m now calling: “Why is Laura Loomer tweeting about obscure deep-cut tech issues as if they are MAGA loyalty tests?”
Last Thursday, a small group of powerful crypto and finance players met at the White House to continue hashing out draft language over stablecoin yields. Coinbase, which sparked these negotiations after it withdrew support from Clarity over stablecoin yields, was in attendance. Prior to the meeting, however, Loomer tweeted a classic banger that demonstrated the tactics she uses to wield influence over Trump: Cast the target as someone who once supported Trump’s enemies and is therefore disloyal.
Ironically, Coinbase has turned into one of the biggest branded boosters of the Trump administration, donating money to his pet initiatives and even having their logo splashed all over last year’s military parade.
Though Loomer tweeted a similar sentiment about Coinbase last June, it seems to have had no impact on whether Coinbase has access to Trump, and likely won’t for a while: I’m told that CEO Brian Armstrong was at Mar-a-Lago the day before Loomer tweeted, attending a World Liberty Financial event.
A wild Trumpworld character has appeared!
If you followed the saga of Logan Paul auctioning off his Pokémon card collection, you may be aware that one of those cards sold for a record-setting $16.5 million last week. But who’s that Pokémon purchaser? It’s AJ Scaramucci, the son of the one and only Anthony Scaramucci, the New York financier and former Trump ally who famously served as Trump’s White House Communications Director in 2017 for 10 days.
AJ is the founder of Solari Capital, which invested $100 million in a Bitcoin mining platform run by Eric Trump. He also now owns the Pikachu Illustrator card, one of only 39 cards in existence and in Grade 10 condition, as well as the diamond chain and carrying case that Paul wore to display the card when he appeared at WrestleMania 38. Scaramucci told reporters that he purchased the card as part of his upcoming “planetary treasure hunt,” adding that he also hoped to purchase a T. rex skull and the Declaration of Independence. (He later posted on X that he hoped to place the card in the Nintendo Museum in Kyoto and cement it as “the ‘Mona Lisa’ of the Pokemon franchise.”)
We can’t believe that a court has to tell you this, much less the Southern District of New York: If you put correspondence between you and your lawyer into a publicly available AI platform, it is no longer protected by attorney-client privilege and becomes subject to discovery!!!!
In any case, have a pleasant State of the Union watch party (if anyone does that anymore) and see you next week.
Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.
Amazon has slowly been teasing out casting details for its live-action adaptation of God of War, and now we have our first look at the show. It’s a single image but a notable one showing protagonist Kratos and his son Atreus. The characters are played by Ryan Hurst and Callum Vinson, respectively, and they look relatively close to their video game counterparts.
There aren’t a lot of other details about the show just yet, but this is Amazon’s official description:
The God of War series storyline follows father and son Kratos and Atreus as they embark on a journey to spread the ashes of their wife and mother, Faye. Through their adventures, Kratos tries to teach his son to be a better god, while Atreus tries to teach his father how to be a better human.
That sounds a lot like the recent soft reboot of the franchise, which started with 2018’s God of War and continued through Ragnarök in 2022. For the Amazon series, Ronald D. Moore, best-known for his work on For All Mankind and Battlestar Galactica, will serve as showrunner. The rest of the cast includes: Mandy Patinkin (Odin), Ed Skrein (Baldur), Max Parker (Heimdall), Ólafur Darri Ólafsson (Thor), Teresa Palmer (Sif), Alastair Duncan (Mimir), Jeff Gulka (Sindri), and Danny Woodburn (Brok).
While production is underway on the God of War series, there’s no word on when it might start streaming.
NEWYou can now listen to Fox News articles!
Your web browser may feel like a safe place, especially when you install helpful tools that promise to make your life easier. But security researchers have uncovered a dangerous campaign in which more than 300,000 people installed Chrome extensions pretending to be artificial intelligence (AI) assistants. Instead of helping, these fake tools secretly collect sensitive information like your emails, passwords and browsing activity.
They used familiar names like ChatGPT, Gemini and AI Assistant. If you use Chrome and have installed any AI-related extension, your personal information may already be exposed. Even worse, some of these malicious extensions are still available today, putting more people at risk without their knowing.
Sign up for my FREE CyberGuy Report
Get my best tech tips, urgent security alerts and exclusive deals delivered straight to your inbox. Plus, you’ll get instant access to my Ultimate Scam Survival Guide – free when you join my CYBERGUY.COM newsletter.
More than 300,000 Chrome users installed fake AI extensions that secretly harvested sensitive data. (Kurt “CyberGuy” Knutsson)
What you need to know about fake AI extensions
Security researchers at browser security company LayerX discovered a large campaign involving 30 malicious Chrome extensions disguised as AI-powered assistants (via BleepingComputer). Together, these extensions were installed more than 300,000 times by unsuspecting users.
Some of the most popular extensions included names like AI Sidebar with 70,000 users, AI Assistant with 60,000 users, ChatGPT Translate with 30,000 users, and Google Gemini with 10,000 users. Another extension called Gemini AI Sidebar had 80,000 users before it was removed.
These extensions were distributed through the official Chrome Web Store, which made them appear legitimate and trustworthy. Even more concerning, researchers found that many of these extensions were connected to the same malicious server, showing they were part of a coordinated effort.
While some extensions have since been removed, others remain available. This means new users could still unknowingly install them and expose their personal data. Here’s the list of the affected extensions:
- AI Assistant
- Llama
- Gemini AI Sidebar
- AI Sidebar
- ChatGPT Sidebar
- Grok
- Asking ChatGPT
- ChatGBT
- Chat Bot GPT
- Grok Chatbot
- Chat With Gemini
- XAI
- Google Gemini
- Ask Gemini
- AI Letter Generator
- AI Message Generator
- AI Translator
- AI For Translation
- AI Cover Letter Generator
- AI Image Generator ChatGPT
- Ai Wallpaper Generator
- Ai Picture Generator
- DeepSeek Download
- AI Email Writer
- Email Generator AI
- DeepSeek Chat
- ChatGPT Picture Generator
- ChatGPT Translate
- AI GPT
- ChatGPT Translation
- ChatGPT for Gmail
FAKE AI CHAT RESULTS ARE SPREADING DANGEROUS MAC MALWARE
These malicious tools were listed in the official Chrome Web Store, making them appear legitimate and trustworthy. (LayerX)
How the fake AI Chrome extension attack works
These fake extensions pretend to offer helpful AI features, such as translating text, summarizing emails, or acting as an AI assistant. But behind the scenes, they quietly monitor what you are doing online.
Once installed, the extension gains permission to view and interact with the websites you visit. This allows it to read the contents of web pages, including login screens where you enter your username and password.
In some cases, the extensions specifically targeted Gmail. They could read your email messages directly from your browser, including emails you received and even drafts you were still writing. This means attackers could access private conversations, financial information and sensitive personal details.
The extensions then sent this information to servers controlled by the attackers. Because they loaded content remotely, the attackers could change their behavior at any time without needing to update the extension.
Some versions could also activate voice features through your browser. This could potentially capture spoken conversations near your device and send transcripts back to the attackers.
If you installed one of these extensions, attackers may already have access to extremely sensitive information. This includes your email content, login credentials, browsing habits and possibly even voice recordings.
We reached out to Google for comment, and a spokesperson told CyberGuy that the company “can confirm that the extensions from this report have all been removed from the Google Web Store.”
BROWSER EXTENSION MALWARE INFECTED 8.8M USERS IN DARKSPECTRE ATTACK
Once installed, the extensions could read emails, capture passwords, monitor browsing activity and send the data to attacker-controlled servers. (Bildquelle/ullstein bild via Getty Images)
7 ways you can protect yourself from malicious Chrome extensions
If you have ever installed an AI-related Chrome extension, taking a few simple precautions now can help protect your accounts and prevent further damage.
1) Remove any suspicious or unused browser extensions
On a Windows PC or Mac, open Chrome and type chrome://extensions into the address bar. Review every extension listed. If you see anything unfamiliar, especially AI assistants you don’t remember installing, click “Remove” immediately. Malicious extensions depend on going unnoticed. Removing them stops further data collection and cuts off the attacker’s access to your information.
2) Change your passwords
If you installed any suspicious extension, assume your passwords may be compromised. Start by changing your email password first, since email controls access to most other accounts. Then update passwords for banking, shopping and social media accounts. This prevents attackers from using stolen credentials to break into your accounts.
3) Use a password manager to create and protect strong passwords
A password manager generates unique, complex passwords for each account and stores them securely. This prevents attackers from accessing multiple accounts if one password is stolen. Password managers also alert you if your login credentials appear in known data breaches, helping you respond quickly and protect your identity. Check out the best expert-reviewed password managers of 2026 at Cyberguy.com.
4) Install strong antivirus software and keep it active
Good antivirus software can detect malicious browser extensions, spyware, and other hidden threats. It scans your system for suspicious activity and blocks harmful programs before they can steal your information. This adds an important layer of protection that works continuously in the background to keep your device safe. Get my picks for the best 2026 antivirus protection winners for your Windows, Mac, Android & iOS devices at Cyberguy.com.
5) Use an identity theft protection service
Identity theft protection services monitor your personal data, including email addresses, financial accounts, and Social Security numbers, for signs of misuse. If criminals try to open accounts or commit fraud using your information, you receive alerts quickly. Early detection allows you to act fast and limit financial and personal damage. See my tips and best picks on how to protect yourself from identity theft at Cyberguy.com.
6) Keep your browser and computer fully updated
Software updates fix security vulnerabilities that attackers exploit. Enable automatic updates for Chrome and your operating system so you always have the latest protections. These updates strengthen your defenses against malicious extensions and prevent attackers from taking advantage of known weaknesses.
7) Use a personal data removal service
Personal data removal services scan data broker websites that collect and sell your personal information. They help remove your data from these sites, reducing what attackers can find and use against you. Less exposed information means fewer opportunities for criminals to target you with scams, identity theft or phishing attacks.
Check out my top picks for data removal services and get a free scan to find out if your personal information is already out on the web by visiting Cyberguy.com.
Get a free scan to find out if your personal information is already out on the web: Cyberguy.com.
Kurt’s key takeaway
Even tools designed to make your life easier can become tools for cybercriminals. Malicious extensions often hide behind trusted names and convincing features, making them difficult to spot. You can significantly reduce your risk by reviewing your browser extensions regularly, removing anything suspicious and using protective tools like password managers and strong antivirus software.
Have you checked your browser extensions recently? Let us know your thoughts by writing to us at Cyberguy.com.
CLICK HERE TO DOWNLOAD THE FOX NEWS APP
Sign up for my FREE CyberGuy Report
Get my best tech tips, urgent security alerts and exclusive deals delivered straight to your inbox. Plus, you’ll get instant access to my Ultimate Scam Survival Guide – free when you join my CYBERGUY.COM newsletter.
Copyright 2026 CyberGuy.com. All rights reserved.
Related Article
Technology
Anthropic refuses Pentagon’s new terms, standing firm on lethal autonomous weapons and mass surveillance
Less than 24 hours before the deadline in an ultimatum issued by the Pentagon, Anthropic has refused the Department of Defense’s demands for unrestricted access to its AI.
It’s the culmination of a dramatic exchange of public statements, social media posts, and behind-the-scenes negotiations, coming down to Defense Secretary Pete Hegseth’s desire to renegotiate all AI labs’ current contracts with the military. But Anthropic, so far, has refused to back down from its two current red lines: no mass surveillance of Americans, and no lethal autonomous weapons (or weapons with license to kill targets with no human oversight whatsoever). OpenAI and xAI had reportedly already agreed to the new terms, while Anthropic’s refusal had led to CEO Dario Amodei being summoned to the White House this week for a meeting with Hegseth himself, in which the Secretary reportedly issued an ultimatum to the CEO to back down by the end of business day on Friday or else.
In a statement late Thursday, Amodei wrote, “I believe deeply in the existential importance of using AI to defend the United States and other democracies, and to defeat our autocratic adversaries. Anthropic has therefore worked proactively to deploy our models to the Department of War and the intelligence community.”
He added that the company has “never raised objections to particular military operations nor attempted to limit use of our technology in an ad hoc manner” but that in a “narrow set of cases, we believe AI can undermine, rather than defend, democratic values” — going on to specifically mention mass domestic surveillance and fully autonomous weapons. (Amodei mentioned that “partial autonomous weapons … are vital to the defense of democracy” and that fully autonomous weapons may eventually “prove critical for our national defense,” but that “today, frontier AI systems are simply not reliable enough to power fully autonomous weapons.” He did not rule out Anthropic acquiescing to the military’s use of fully autonomous weapons in the future but mentioned that they were not ready now.)
The Pentagon had already reportedly asked major defense contractors to assess their dependence on Anthropic’s Claude, which could be seen as the first step to designating the company a “supply chain risk” – a public threat that the Pentagon had made recently (and a classification usually reserved for threats to national security). The Pentagon was also reportedly considering invoking the Defense Production Act to make Anthropic comply.
Amodei wrote in his statement that the Pentagon’s “threats do not change our position: we cannot in good conscience accede to their request.” He also wrote that “should the Department choose to offboard Anthropic, we will work to enable a smooth transition to another provider, avoiding any disruption to ongoing military planning, operations, or other critical missions. Our models will be available on the expansive terms we have proposed for as long as required.”
Technology7 minutes ago
Here’s your first look at Kratos in Amazon’s God of War show
World13 minutes ago
State Dept authorizes non-essential US Embassy personnel in Jerusalem to depart ahead of possible Iran strikes
Politics19 minutes ago
Where Iran’s ballistic missiles can reach — and how close they are to the US
Health25 minutes ago
Heart disease threat projected to climb sharply for key demographic
Sports31 minutes ago
Anthony Richardson free to seek trade after injury setbacks amid Colts’ shift to Daniel Jones
Florida1 year ago
Florida High School Football Rankings: Top 25 teams – Oct. 21
Connecticut2 years ago
Cleary's 21 help Le Moyne down Central Connecticut State 69-64 in OT
Oregon2 years ago
How old is Bo Nix? What to know about Oregon quarterback ahead of 2024 NFL Draft
Virginia2 years ago
99th annual Pony Swim held in Virginia
Indiana1 year ago
Indiana Members Credit Union announced as new anchor tenant at Bottleworks District
Politics19 minutes ago
Where Iran’s ballistic missiles can reach — and how close they are to the US
World1 hour ago
Has India’s influence in Afghanistan grown under the Taliban?
News2 hours ago
Bill Clinton to testify before House committee investigating Epstein links
News11 hours ago
Read Judge Schiltz’s Order
Politics12 hours ago
Vulnerable House Dem lashes out at Trump’s ‘racist’ SOTU challenge: ‘That was uncomfortable’
-
World1 day agoExclusive: DeepSeek withholds latest AI model from US chipmakers including Nvidia, sources say
-
Massachusetts2 days agoMother and daughter injured in Taunton house explosion
-
Montana1 week ago2026 MHSA Montana Wrestling State Championship Brackets And Results – FloWrestling
-
Oklahoma1 week agoWildfires rage in Oklahoma as thousands urged to evacuate a small city
-
Louisiana4 days agoWildfire near Gum Swamp Road in Livingston Parish now under control; more than 200 acres burned
-
Technology6 days agoYouTube TV billing scam emails are hitting inboxes
-
Denver, CO2 days ago10 acres charred, 5 injured in Thornton grass fire, evacuation orders lifted
-
Technology6 days agoStellantis is in a crisis of its own making