Tag: big tech

Meta putting up tents across the US to house AI servers, like ‘a scene out of the movie Mad Max’ — structures take three months to build and use jet engines for power

Meta has moved from building traditional structures for its data centers to putting up tents across the U.S. and sticking AI servers inside them. Michael Thomas, founder of market intelligence and data center tracking firm Cleanview Energy, said on X that the AI tech firm has already built or is in the process of constructing three data centers that use the strategy.

One site, located in New Albany, Ohio, already has five buildings that took approximately two to three years to complete. The company then started putting up five tents, with an area of around 125,000 square feet each, in the area. City permits seen by Cleanview Energy say that the construction started in April 2026, while recent satellite images show that the structures have already been completed.

Meta CEO Mark Zuckerberg first announced the strategy of pitching tents and filling them with AI servers last year. It seems that he wanted the infrastructure to come online quickly while demand for compute is increasing exponentially. It’s said that Meta is inspired by Elon Musk’s feat with xAI, which built a 100,000-strong AI data center in just 19 days in 2024 — something which usually takes four years, according to Nvidia CEO Jensen Huang. The technique is apparently quite effective, and it’s now being applied to two other sites, including one in Tennessee.

Putting AI servers inside tents, officially called “rapid deployment structures,” is one of the more unique approaches to the AI build-out, Thomas said. They’re certainly not as sturdy as physical buildings made from steel and concrete, with one commenter comparing it to the “classic $10k racing bike with a $9 lock” situation. Nevertheless, the company has probably weighed the pros and cons of such a setup and has decided that it was worth taking the risk to gain an advantage in the AI infrastructure race.

Another factor that allowed Meta to bring its data centers online at a much faster pace is its use of “behind-the-meter” power, in which the company installed its own turbines to produce power on-site rather than relying on grid power. This is similar to what Musk did with his Memphis Supercluster, which he initially powered with portable power generators. However, Meta’s turbines would be a permanent feature on the Ohio site, as it’s designed to run independently of the power grid.

Keep reading

Five tech giants are hiding $1.65tn in AI debt, using the trick that toppled Enron

Look at what Alphabet, Microsoft, Amazon, Meta, and Oracle officially owe, and the numbers seem large but manageable. Look off the books, and a second, bigger pile of debt appears.

A Nikkei study put that hidden figure at $1.65 trillion, up roughly eightfold in four years. It is more than the $1.35 trillion the five report outright.

The Enron echo

The money is tied up in off-balance-sheet vehicles, the same kind of structure Enron used to hide debt before it collapsed 25 years ago. Back then it was fraud. Now, tightened rules and fuller disclosures make it legal.

The tools are still there, though. “Enron’s crime wasn’t having special purpose vehicles,” analyst Gil Luria told Bloomberg Law. “Enron’s crime was hiding them.”

The mechanics are simple. A company packages the debt for chips, servers, and power into a separate legal entity, often a joint venture, so the cost never flows through its own accounts.

Take Meta’s Hyperion data centre in Louisiana. Meta and Blue Owl Capital both put equity into a separate structure that took on $27 billion in debt. Meta is the sole tenant, yet argues it does not have to record that debt, because it is not the one who must find replacement tenants.

Oracle has $260 billion of future lease commitments that will eventually land on its books. Nvidia carries $119 billion in purchase obligations. Alphabet and Microsoft keep their vehicles off-book too.

The numbers behind the boom

The scale is the story. Meta’s off-balance-sheet debt alone is about $420 billion, nearly triple its reported debt. Oracle’s has grown roughly thirtyfold in four years.

It is all in service of the same race. The industry is expected to spend more than $3 trillion through 2028 building and equipping AI data centres, much of it financed against the chips inside them.

Why it matters now

The timing is awkward. Four of the five report earnings in the next two weeks, and the reported debt will look tidy. The $1.65 trillion sitting in the footnotes will not make the headlines.

The catch is what happens later. When a data centre goes live, its lease rolls onto the balance sheet at once. If AI demand falls short, the facility is marked down, and the loss lands on the lenders and insurers who funded it.

Some already see the risk. S&P has cut Oracle’s credit rating over stretched leverage, and both Morgan Stanley and Moody’s have flagged the wider issue. “What if one of these companies was a house of cards,” asked accounting consultant Tom Selling, “and was propping itself up with this accounting treatment?”

Keep reading

Google DeepMind CEO Calls for U.S.-Led Global AI Safety Watchdog

Demis Hassabis, the co-founder and CEO of Google DeepMind, is advocating for the United States to create a new artificial intelligence oversight body with authority to evaluate the world’s most sophisticated AI models and potentially coordinate industry-wide slowdowns when risks escalate.

Axios reports that the Nobel Prize-winning scientist behind Google’s ultra-woke Gemini AI system outlined his proposal in a personal manifesto published this week titled “A Framework for Frontier AI and the Dawning of a New Age.” In an exclusive interview with Axios, Hassabis emphasized the urgency of implementing a more systematic regulatory approach to artificial intelligence, one that would be industry-funded, staffed by leading technical experts, and accountable to the U.S. government.

Speaking from his London headquarters, Hassabis characterized current AI-driven cybersecurity risks as warning signals of greater dangers ahead. He predicted that within 18 months, these capabilities, along with potentially catastrophic biological and nuclear threats, could exist within open-source AI models that would be impossible for any government to control. The DeepMind CEO stressed that risks would emerge not only from open-source models but also from the more powerful proprietary systems being developed by major AI laboratories.

“What we collectively do now will determine how the next phase of civilization unfolds,” Hassabis wrote in his manifesto.

Hassabis has spent recent months conducting private consultations to build support for his proposal, meeting with Trump administration officials, leaders of other AI laboratories, and European government representatives before making his plan public. He reported receiving positive responses from the administration, which had previously adopted a hands-off stance toward AI regulation before the recent Mythos incident raised alarm bells.

The DeepMind chief, who commands significant respect across different factions within the AI community, indicated that leaders of other major AI labs have expressed agreement with the general direction of his proposal. “This is where the industry needs to go,” Hassabis said of the feedback he has received. His timeline for implementation is ambitious, aiming for the new regulatory body to become operational before the end of the current year.

Keep reading

Zuckerberg to spend over $10B on “historic” Alberta AI data centre investment, sources say

Meta Platforms, the parent company of Facebook, Instagram and WhatsApp, is behind a massive artificial intelligence data centre planned for Sturgeon County, Alberta, Juno News has confirmed.

This is according to several well-placed sources with direct knowledge of the investment, one that all sources agree will be “historic” in magnitude.

The project is expected to involve roughly $13 billion in total investment, though the final figure could still change as it is unclear if the final proposal has been approved and signed off by Mark Zuckerberg and Meta’s board.

Keep reading

Anthropic Removes “Scary” Secret Claude Tracker After Developer Stumbles Across It

Anthropic has removed hidden detection code from its Claude Code tool after a developer reverse-engineered the binary and exposed how the company was subtly monitoring users in China.

The code, which Anthropic described as an experiment launched in March, used a form of prompt steganography to signal information about a user’s environment back to Anthropic’s servers. It was designed to help detect unauthorized resellers and attempts by other organizations to distill Claude’s capabilities into their own models.

How The Detection Worked

The mechanism was first spotted by a Reddit user known as LegitMichel777, who stumbled on it while trying to restore a disabled feature in Claude Code. A separate developer known as Thereallo independently confirmed the finding the same day, June 30, publishing a technical breakdown of exactly how it worked.

The checks only ran in one specific situation: when a user pointed Claude Code at a different server instead of Anthropic’s own – something companies commonly do when they route their traffic through internal systems or third-party gateways. From there, it checked two things:

  • Whether the user’s computer was set to a Chinese time zone (Shanghai or Urumqi).
  • Whether the new server address matched a hidden list of Chinese AI companies (including well-known names like DeepSeek, Zhipu, and Moonshot) or known resale and proxy services.

If either check came back positive, Claude Code would quietly tweak a line of text called the “system prompt” – background instructions the app automatically sends to the AI model with every request, invisible to the person typing. Specifically, it changed how the date was written in that line:

  • If the user was in a Chinese time zone, the date switched from using dashes to slashes (e.g., 2026/06/30 instead of 2026-06-30).
  • The apostrophe in the phrase “Today’s date is…” was swapped for one of three lookalike characters, each one a different signal, depending on which combination of checks the session had triggered.

None of this was visible to users, or likely even to the AI model itself in normal use – the characters look identical on screen. But Anthropic’s servers could read the difference instantly. The lists of flagged domains and keywords were also scrambled inside the app’s code using a basic encryption trick, so they wouldn’t show up if someone just opened the file and searched for them.

Thereallo called the approach “prompt steganography” – hiding a signal inside ordinary-looking text – and noted it let Anthropic sort and flag sessions without needing any separate, visible tracking system.

Anthropic’s Explanation

Last Tuesday, Anthropic engineer Thariq Shihipar, who works on the Claude Code team, confirmed the feature on X:

“This is an experiment we launched in March that was meant to prevent account abuse from unauthorized resellers and protect against distillation. The team has landed stronger mitigations since then and we’ve actually been meaning to take this down for a while. We merged the PR and this should be fully rolled back in tomorrow’s release.”

Anthropic has stated that unauthorized resellers have been selling access to Claude accounts and subscriptions at steep discounts in certain markets. The company has also publicly documented large-scale efforts by Chinese AI labs to distill its models by querying them at high volume through proxies and fraudulent accounts.

Anthropic removed the detection logic shortly after it became public.

Keep reading

The AI Jobs Apocalypse Has Begun: We Are Witnessing An Unprecedented Wave Of AI-Related Layoffs In 2026

We all knew that this would be coming. In 2026, the rate of AI-related layoffs has greatly accelerated, and that means that large numbers of good paying jobs are suddenly disappearing from the economy. This is occurring when we are already facing one major crisis after another, and so the timing could not be worse. Many young people specifically chose a major in college that would prepare them for positions in the tech industry because those were supposed to be the jobs of the future. Unfortunately, the AI jobs apocalypse is wiping out those jobs the fastest.

Let me give you some cold, hard numbers that will clearly demonstrate what I am talking about.

According to Challenger, Gray & Christmas, the number of announced job cuts in the United States last month was the highest that we have seen during the month of May since the peak of the last pandemic…

U.S.-based employers announced 97,006 job cuts in May, up 16% from the 83,387 job cuts recorded in April, and up 3% from the 93,816 announced in the same month last year, according to a report released Thursday from global outplacement and executive coaching firm Challenger, Gray & Christmas.

May’s total is the highest for the month since 2020, when 397,016 job cuts were recorded in May at the height of the pandemic. It also marks the third straight month that cuts have risen, climbing from 48,307 in February to 97,006 in May.

There is no way to spin those numbers to make them look good.

And for the third month in a row, artificial intelligence accounted for more layoffs than any other reason…

In May, Artificial Intelligence (AI) led all reasons for job cuts for the third month in a row, with 38,579 announced cuts. It is the highest monthly total ever recorded for the reason since Challenger began tracking it in 2023, and it accounted for 40% of all cuts announced in May — up from just 7% in January, 25% in March, and 26% in April. For the year, AI has been cited in 87,714 cuts, or 22% of all 2026 layoffs, already far surpassing the 54,836 attributed to the reason in all of 2025.

Read that last sentence again.

The number of AI-related layoffs in 2026 has already surpassed the grand total of AI-related layoffs for the entire year of 2025.

That is how fast things are now moving.

Needless to say, the tech industry is being hit the hardest.

At this point, tech layoffs are running 44 percent faster than last year…

So far this year, there have been an estimated 363 layoffs at tech companies this year, affecting nearly 150,000 people — a pace of about 974 people per day, 44% faster than last year — according to TrueUp, a tech job board and recruiting platform that also runs one of the most widely cited tech layoff trackers.

The trend appears to be accelerating. Tech layoffs hit their highest single month in two years last month, with nearly 40,000 cuts, and AI was the most-cited reason for layoffs across every industry for the third month running, according to outplacement firm Challenger, Gray & Christmas.

There is no long-term loyalty in the tech industry anymore.

Keep reading

Supreme Court Rules Police Conduct a Fourth Amendment “Search” When Grabbing Your Google Location History Data Through Geofence Warrants

The U.S. Supreme Court held Monday that law enforcement officers conduct a Fourth Amendment search when they obtain cell phone users’ precise Location History data from Google using a geofence warrant.

In a 6-3 decision in Chatrie v. United States, the Court ruled that Americans have a reasonable expectation of privacy in their cell phone location information, even when that data is stored by a third-party technology company such as Google. The ruling represents one of the Court’s most significant digital privacy decisions since its 2018 Carpenter decision involving historical cell-site location data.

Justice Elena Kagan authored the majority opinion, joined by Chief Justice John Roberts and Justices Sonia Sotomayor, Brett Kavanaugh, Ketanji Brown Jackson, and Jackson separately concurring.

Justice Neil Gorsuch concurred only in the judgment, while Justice Samuel Alito dissented, joined in part by Justices Clarence Thomas and Amy Coney Barrett. Justice Barrett also filed a separate dissent.

This builds directly on the landmark Carpenter v. United States (2018) decision, which already required warrants for cell-site location information (CSLI).

The Court made clear that Google’s even more precise and sweeping Location History data — which logs a user’s location every two minutes or so, within about 20 meters, and can even reveal elevation and which floor of a building someone is on — deserves at least the same protection.

The case, Chatrie v. United States (No. 25-112), arose from a May 20, 2019, armed robbery of a credit union in Midlothian, Virginia. Police had surveillance footage and witness statements but no suspect. On June 14, they obtained a Virginia magistrate’s geofence warrant directed at Google.

Keep reading

Meta Restricts Engineers’ Use of Claude Code And Codex Over Model ‘Distillation’ Concerns

Meta Platforms has instructed engineers in its Applied AI division to limit or restrict their use of Anthropic’s Claude Code and OpenAI’s Codex coding and agent tools, according to internal documents reviewed by The Information. The policy, driven by concerns over inadvertent model distillation, aims to prevent outputs from rival AI systems from contaminating Meta’s own training data and model development processes for its Llama family of models (which, quite frankly, could only help).

The move reflects the increasingly zero-sum nature of frontier AI development, where companies aggressively protect the provenance and purity of their training data while seeking to reduce reliance on competitor tools. Internal guidelines referencing the restrictions date back to at least May, with the policy actively in effect as of late June. Meta has not publicly confirmed or commented on the directive.

According to the internal documents, strict limits have been placed on how engineers in the applied AI division can use the rival tools. The stated goal is to block “inadvertent distillation” of competitor model outputs into Meta’s AI development pipeline. The scope is targeted: it focuses on engineers working directly on model building and applied AI initiatives rather than the entire engineering organization.

Claude Code from Anthropic and Codex from OpenAI are basically the industry standard now for professional developers engaged in agentic coding workflows. These desktop and app-based interfaces can plan, write, debug, and iterate on complex codebases, offering powerful assistance at relatively low individual subscription costs. That accessibility, however, has increased the potential surface area for the risks Meta is now seeking to contain.

Keep reading

Meta pauses an AI training program that tracks employees’ keystrokes after an internal leak

Meta is pausing an internal AI training program after sensitive data was accessible across the entire company, according to screenshots obtained by Business Insider.

A screenshot showed that the leak exposed employees’ private conversations, performance data, and transcriptions. The incident was classified as a SEV 2 on a scale of 0 to 5, with 0 being the most severe.

A Meta spokesperson confirmed the incident and said the company is investigating.

“We have carefully designed this program with privacy safeguards, and while we have no indication at this time that any data was improperly accessed by Meta employees, we’re pausing it while we investigate,” the spokesperson said.

In April, Meta announced the AI training program, called the Model Capability Initiative (MCI), which was intended to improve the company’s AI models by using its staff’s keystrokes and mouse movements as training data. The program, which is mandatory for most staff, sparked a backlash from employees who felt uncomfortable with their data being recorded, Business Insider previously reported.

This leak is causing frustration within Meta, according to screenshots seen by Business Insider, with employees critical that data wasn’t locked down from the start.

“I am incensed,” one employee wrote on Monday about the recent leak in an internal group, according to a screenshot obtained by Business Insider.

Keep reading

No New Laws Required… Private Biometrics Are Building The Digital ID Prison

That “black pill moment” is arriving faster than many realize. Not primarily through sweeping new government mandates, but through private companies quietly normalizing biometric data collection under the banners of “security,” “fraud prevention,” and “child protection.” They are erecting the infrastructure for a world where you cannot easily participate in daily life, commerce, or even basic online access without surrendering your face, your license scan, or other biometrics. Once the systems exist and the data flows, laws can simply ratify what private actors have already made routine.

In a recent commentary “Digital ID Black Pill Moment”, I highlighted a sobering reality: 186 out of 198 countries already have digital ID systems in place. Only a shrinking handful of nations lack foundational national digital IDs. As I wrote, “the global push for digital IDs is far advanced, likely past the point of no return, aligning with the UN’s 2030 goal of universal legal identity and enabling a globalist digital currency system that could control access to everything.”

Facebook/Meta: Selfie or Stay Locked Out

Government mandates are not required to finish building the digital surveillance prison. Citizens are willingly submitting their biometrics to access social media sites. For example, I am no longer on Facebook. They banned me during the Covid era after I began sharing information about the true contents of the shots and alternative treatments. A friend just sent me a Facebook post and I could not view it without taking a selfie and sending it to FB. No way was I going to comply.

Try viewing certain Facebook posts or recovering a flagged account, and you may hit this wall. Users are increasingly prompted to submit a video selfie turning their head in different directions so the system can map facial geometry to “prove you’re a real person” or restore access. The company states it uses this to combat scams and compromised accounts, and claims the video is deleted after verification.

Keep reading