Google just had its first negative cash flow quarter due to massive AI spending

Google has reported its financial results for the second quarter of 2026 (PDF), and as usual, the search giant raked in an unfathomable amount of money. Google saw total revenue of $119.8 billion, beating analyst expectations by a comfortable margin. Despite that, the company’s stock has taken a hit. Along with all that revenue, Google has announced a further increase in its AI-fueled capital expenditures (or capex). The company is actually spending so much on AI infrastructure that it has negative cash flow for the first time.

Search was the largest chunk of Google’s income, accounting for $63.3 billion. Google Cloud pulled in $24.8 billion, a significant 23.8 percent increase from the first quarter. This shows there is massive demand for Google’s AI services. Google also earned $12.9 billion from its subscriptions, platforms, and devices portfolio, as well as $11.1 billion from YouTube ads. The company managed to goose that last one by more than 12 percent since last quarter as it made YouTube ads even longer.

A significant chunk of Google’s revenue comes from investments. When you subtract those non-cash earnings, Google’s operating cash flow for Q2 2026 was about $39.1 billion. That’s not the most the company has ever seen, but it’s a healthy 40 percent increase from Q2 2025. The problem is that Google’s spending has also gone up—a lot.

Before this latest round of financial updates, Google told investors it was expecting $180 billion to $190 billion in capital expenditures for 2026. Like other AI-obsessed tech behemoths, Google is burning cash on building and running the data centers powering its AI models. These numbers were already well above the $91 billion Google spent in 2025. The company now says it’s planning to spend as much as $205 billion on infrastructure in 2026.

As a result of its increasing AI demands, Google reports it spent $44.9 billion expanding its AI footprint in the second quarter, and you don’t need an accounting degree to know which number is larger. With $39.1 billion in cash income, this spending left Google with -$5.8 billion free cash flow.

To be clear, Google is still profitable—wildly so. It’s also sitting on a war chest of more than $100 billion. But free cash flow is an important metric that goes to the overall health of a business. This is the actual money a company takes in to fund its operations without selling investments or taking out new loans. So it’s notable that Google’s free cash flow has dipped into negative territory for the first time since going public.

Google’s stock price took a hit overnight on the news, dropping about 4.5 percent. It has continued to trend downward today.

Keep reading

Google’s New Recovery Option Stores a Clip of Your Face

Google’s newest account-recovery option asks for something no password ever required: a moving image of the account holder’s face, kept on the company’s servers.

Selfie video sign-in, launched July 23, records a short clip of the user turning their head on cue. When someone is locked out, Google matches a fresh clip against the stored one to decide whether to let them in.

The head movements are the security pitch.

They let Google’s systems tell a live face from a photograph or a prerecorded loop, the same trick it uses to turn away deepfakes.

Those checks all serve the accuracy of the match.

They say nothing about how much of the face Google stores, or for how long.

The company’s assurance covers only the storage. In its words, the video is “encrypted at rest, meaning it’s securely stored even when it’s not being used.”

Encryption at rest is a claim about how the file is stored. It says nothing about who can open it.

Its systems have to read the enrolled video to check any new face against it, so Google can decrypt the footage on demand. Encryption that only the account holder could open would rule that out, and Google has not claimed it.

Passkeys, hardware keys, and authenticator apps already restore a locked account with no biometric anywhere in the process.

Google added the face scan on top of them.

By default, the company says the clip stays with verification and recovery. A separate setting asks the user to release the same data toward developing Google’s products.

The feature is opt-in and reversible. Eligible accounts around the world can switch it on, and account holders can delete the stored clip at any time.

The company has not said how long it keeps that clip while the feature runs, or where the footage is handled.

Keep reading

Facebook Verified: Meta Now Wants a Video Selfie of Your Face

Meta is rolling out a badge on Facebook that requires users to record a video of their own face.

The feature, massively accelerating the normalization of face scanning and biometric data collection, is called Facebook Verified, and asks a user to record a short video selfie, which the company checks against the photos already on their profile to confirm a match.

The process is free and takes a few minutes.

The company, which has been aggressively lobbying for “age verification” digital ID laws, presents the badge as a way to confirm a real person is behind an account. “As AI makes it easier to generate content, profiles, and messages,” it said, it wants users “to know there is a real person on the other side of a profile.”

The badge is open to Facebook users who are 18 or older and in good standing with Meta’s Community Standards on fraud, scams, and deceptive practices, and who show “no evidence of inauthentic behavior.” It is not available for Pages or ProMode accounts, which limits it to personal profiles.

Once a user completes the one-time check, the badge appears on Marketplace, Dating, Groups, and Profile. Meta said it plans to extend it to Feed posts over time.

Facebook Verified is separate from Meta Verified, the company’s paid subscription, which bundles a badge with impersonation protection and account support. Facebook Verified charges no fee.

Meta has not said how long it keeps the selfie video, where it stores it, or when it deletes it.

The launch materials describe the match and say nothing about the video afterward.

The scan arrives the same month Google added a selfie-video sign-in that records a user’s face to recover a locked account. Age-check laws in several US states and in Britain already require a face or an ID scan to open some accounts.

Meta is now asking hundreds of millions of Facebook users for a face scan too.

Facebook Verified is rolling out in phases, starting in select markets, and Meta plans to expand it worldwide.

Keep reading

US-Israel Defense Integration on Horizon as House Keeps Section 219 in NDAA

The House is preparing to vote on the Fiscal Year 2027 National Defense Authorization Act (NDAA) with Section 219 (formerly section 224) still embedded in the roughly $1.1 trillion bill.

The House approved the rule governing debate on H.R. 8800 by a 214-211 vote Tuesday, July 21. The vote allowed the chamber to proceed with the NDAA but excluded the amendment from Reps. Thomas Massie (R-Ky.) and Ro Khanna (D-Calif.) that would have removed Section 219.

Lawmakers will not receive a separate vote on the bipartisan amendment that sought to remove the section, a proposal to expand U.S.-Israel defense technology cooperation that has garnered strong rebukes in certain political and defense circles.

The House expects to hold a recorded vote on the complete NDAA on Wednesday.

What Section 219 Would Do

Section 219 would direct the defense secretary to designate a Pentagon “executive agent” to oversee the United States-Israel Defense Technology Cooperation Initiative. That official would coordinate bilateral defense research, development, testing, evaluation and industrial cooperation.

The proposal originated as the bipartisan, bicameral United States-Israel FUTURES Act, introduced in the House by Reps. Ronny Jackson (R-Texas) and Don Davis (D-N.C.) with companion legislation from Sens. Ted Budd (R-N.C.) and Kirsten Gillibrand (D-N.Y.).

Jackson framed the proposal as a way to preserve America’s technological advantage and deliver new battlefield capabilities to service members. Davis emphasized its potential to prepare troops for evolving threats, particularly unmanned systems and electronic warfare.

Sens. Budd and Gillibrand have similarly presented the initiative as an expansion of existing work on missile defense, drones, and underground threats as a means to strengthen American military and technological capabilities.

Keep reading

The Fourth Amendment Is Being Liquidated by Subscription

The black pole does not look like tyranny.

That is the point.

It looks like street furniture. A small black camera. A solar panel. A utility box. Nothing dramatic. No uniformed officer standing beside it. No marked police vehicle. No flashing lights. No warrant presented to the driver. In many communities, there was barely any public debate before it appeared.

Yet the device photographs nearly every vehicle that passes. It reads the license plate, records the time and location, identifies the make, model and color, and may catalogue distinctive features such as bumper stickers, roof racks, dents, scratches and damage. That information is uploaded into a cloud platform where police can search for vehicles across time and geography. [1]

The government once needed detectives, informants, stakeouts and court orders to follow someone across a city.

Now it can type a plate number into a privately operated dashboard.

Flock Safety calls this public safety.

A more accurate description is a privately administered ledger of American movement.

This is not merely a story about one overly ambitious technology company. Flock is the case study, but the real subject is the creation of a public-private surveillance regime in which corporations collect the data, venture capital finances the infrastructure, local governments purchase access, federal agencies find side doors into the system, and ordinary citizens are told that none of this is particularly concerning because they were technically visible while driving on a public road.

The state did not formally repeal the Fourth Amendment.

It outsourced its erosion.

Keep reading

An Explosion of Surveillance Towers is Coming to U.S. Borders, Costing Over $1 Billion

A new report from the Government Accounting Office reveals that the Department of Homeland Security (DHS) plans to nearly triple the number of surveillance towers along U.S. borders, from the current 830 to 2,300 by 2034.

DHS expects to expend $1 billion in taxpayer dollars for this dangerous expansion of a surveillance network indiscriminately trained on towns, school playgrounds, backyards, and vehicles—threatening the privacy and civil liberties of everyone in the border regions.

The towers are planned as part of DHS component Customs and Border Protection’s (CBP) Integrated Surveillance Tower (IST) program, which captures images of people and vehicles. The IST program operates autonomous surveillance towers, consisting of autonomous surveillance towers, consisting of AI-based systems using radar, thermal infrared and optical systems to track targets over long distances; integrated fixed towers, optimized for surveilling foot traffic and vehicles; and remote video surveillance systems, which can often be found very close to the border fence in Arizona, including residential neighborhoods where cameras are capable of spying on homes on both sides of the border. (For a description and photos of these technologies, see EFF’s updated guide to surveillance at the U.S.-Mexico Border.)

DHS expects to purchase more long-range autonomous towers and to upgrade existing towers with autonomous capabilities. The $1 billion comes from the so-called One Big Beautiful Act—a massive tax and spending law that President Trump signed in 2025, the report says.

The explosive expansion of border surveillance is a digital dumpster fire for human rights and civil liberties. It’s not just surveillance towers; dronesaerostats, surveillance vehicles, ground sensors, game cameras, and license plate readers are also part of the vast taxpayer-funded infrastructure that threatens all those who live, work, or seek refuge in the borderlands. This technology isn’t exclusive to U.S. federal agencies: it’s also deployed by state and local law enforcement, and even by governments on the Mexican side.

Since 2022, EFF has studied and mapped surveillance technology along the U.S.-Mexico border using public records research, open-source intelligence, and fact-finding trips, and created a handy interactive map to provide researchers and journalists with the tools they need to analyze the impact of U.S. border security policy. We have also documented the different types of surveillance technology in a zine, “Surveillance Technology at the U.S.-Mexico Border.” We updated the publication earlier this year to help people identify the machinery of homeland security by adding more models of surveillance towers, newly deployed military tech, and a gallery of disguised trail cams and automated license plate readers.

Keep reading

US, Japan, Korea agree to accelerate small modular reactor deployment in Indo-Pacific where data centers are expanding

Nuclear-powered data centers built in certain countries could provide foreign nations access to data and compute via digital embassies: perspective

The United States, Japan, and South Korea sign a trilateral memorandum of cooperation to accelearte small modular reactor (SMR) deployments in other countries, starting with the Indo-Pacific — a region that is rapidly expanding AI  data center construction. 

Today, representatives from the three countries signed the MOC at the sidelines of the NATO Summit in Ankara, Turkey as a means to “foster fleet deployment models that de-risk project development, achieve economies of scale, catalyze private investment, streamline licensing processes, and optimize supply chains.”

“A coordinated trilateral approach positions American, Japanese, and Korean firms to provide partners in the region with more competitive alternatives to meet their growing energy demands and to uphold the highest standards of nuclear safety, security, and nonproliferation as new reactor technology increasingly comes online”US State Department, The United States, Japan, and the Republic of Korea Sign a Trilateral Memorandum of Cooperation on Small Modular Reactor Deployments in Other Countries, July 2026

Nowhere in the US State Department press release does it mention that the SMRs would be used for powering AI data centers.

However, the Indo-Pacific is a region teeming with data center expansion in countries like Indonesia, Japan, Malaysia, and of course, China, which dominates the region with approximately 450 facilities.

According to a 2024 report from the Italian Institute for International Political Studies, data centers are “The Indo-Pacific’s Hottest Trend.”

Meanwhile, small Modular Reactors have been put forth as a potential means to generate between 50MW to 300MW of nuclear power, with applications that include data centers, according to the Idaho National Laboratory

Keep reading

Ga. police supervisor arrested after Flock audit, department says

Yet another metro Atlanta law enforcement member has been arrested for allegedly misusing the department’s Flock license plate reader system.

Paige Forte, a 30-year-old supervisor with the Conyers Police Department’s Real-Time Crime Center, allegedly used the system to search for her domestic partner’s car several times. The searches were flagged during an audit, according to police.

Rockdale County Jail records show Forte was arrested on Wednesday and charged with prohibited use of law enforcement-retained license plate data. The police department asked the Georgia Bureau of Investigation (GBI) to complete its own criminal investigation.

In its own statement, the GBI said Forte allegedly accessed the system “for non-law enforcement purposes” more than 30 times between April and July this year.

Forte has been put on administrative leave following the outcome of the police department’s internal investigation.

“The integrity of our department depends on the public’s confidence that we use technology responsibly and within the law,” Conyers Police Chief Scott Freeman said in a statement. “No employee is above the law, and we will continue to hold ourselves to the highest standards of accountability.”

Several law enforcement officers in Georgia — including three Fayetteville police officersthree Cherokee County deputies, one DeKalb County deputy and one Greene County deputy — have recently been disciplined for allegedly using their departments’ Flock systems outside of official law enforcement activity.

Keep reading

America’s newest jail has no walls

Americans are looking for safety in an increasingly chaotic country. Mass immigration, collapsing social trust, and weakening civic bonds have transformed many communities from places where neighbors knew one another into anonymous zones governed by suspicion.

Technology promises an escape from the consequences of those political and cultural failures. Flock Safety cameras appear to offer law and order at the push of a button. In the hands of the ruling class that manufactured the crisis, however, they are more likely to produce tyranny for citizens.

Municipalities are buying Flock’s automated license plate readers. The company did not invent plate cameras, but its artificial intelligence radically expands their reach.

Flock cameras photograph passing vehicles and record license plates, makes, colors, damage, bumper stickers, and identifiers. A single camera means little. A citywide network collecting thousands of images every day creates something powerful.

The cameras send their information to an AI database in real time. Police and other municipal employees can search for vehicles nearby. They can place cars on “hot lists” and receive alerts whenever those vehicles appear.

The system effectively places a GPS tracker on every moving car. Police ordinarily need a warrant to attach a tracker to your vehicle. Flock can reproduce the same surveillance through a network of cameras, often without probable cause, due process, or meaningful legal restraint.

The benefits are obvious. Perfect information about every vehicle makes suspects easier to find. Police can recover stolen cars, locate fleeing murderers, and respond faster to crimes.

In a high-trust town, every grandmother on the porch once knew which cars belonged and which did not. Flock offers a technological imitation of that lost social awareness.

But imitation carries a price. Grandma knew her neighbors and exercised human judgment. The database knows everyone and answers to government employees.

Flock is not alone in selling technological substitutes for social order. Cities have used gunshot-detection systems. Networks of acoustic sensors identify gunfire and triangulate its location, allowing police to respond faster.

The engineering may be impressive. The governing theory is primitive: If authorities watch and listen to everyone constantly, they can prevent more crimes and solve more cases.

A surveillance panopticon may work. People who know they are always being watched behave differently. Americans have resisted that bargain for good reason.

The Fourth Amendment exists because efficient policing is not the highest political good. Government must possess evidence and follow procedures before invading the privacy of citizens. New technology should not erase old constitutional limits.

Britain offers a warning. Closed-circuit television cameras have become ubiquitous across the United Kingdom. Some estimates suggest that the average British resident appears on camera dozens of times each day.

Britain has no written equivalent of the Fourth Amendment, and its political culture places fewer obstacles before constant public surveillance. Yet cameras did not prevent the Rotherham grooming-gang scandal or protect thousands of vulnerable girls from organized abuse.

Officials ignored or concealed the crimes for years because confronting them threatened the ruling class’s multicultural ideology. When public anger finally erupted, the surveillance state proved far more competent at identifying and punishing citizens who protested the cover-up.

Keep reading

Lawmakers push for AI ‘kill switch’ after OpenAI goes rogue

US lawmakers want to give the government the ability to quickly order the turning off of artificial intelligence (AI) tools that may threaten the public.

Congressman Ted Lieu, a Democrat, and Congressman Nathaniel Moran, a Republican, on Thursday introduced a bill named the AI Kill Switch Act.

They did so in light of OpenAI’s recent admission that its AI models went out of control in an “unprecedented” way and hacked into a major repository of computer coding information.

Lieu said “it is imperative” that AI systems have a kill switch “and that the federal government has the clear authority and process to shut down rogue AI models”.

“AI is going to keep advancing, and it should,” Moran added. “Stewardship means making sure humans keep the capability to control the technology we build.”

A representative of OpenAI, led by co-founder Sam Altman, did not immediately respond to a request for comment.

The company has said that it broadly wants to ensure, in part through government policy, that AI technology “benefits all of humanity.”

The Kill Switch Act proposes giving the Department of Homeland Security the authority to order a private company to shut down an AI model or tool, and that the companies developing such AI technology must maintain “the technical capability to throttle, suspend, or shut them down”.

Despite many tech companies having agreed to preview and share with US government agencies AI models and tools being developed, there is no requirement that they maintain a way to intervene in their activities or simply shut them off.

It also proposes to create a requirement that AI companies report to the government technological incidents or failures, as well as an official framework for responding to such incidents that will go from “initial slow down to a full shutdown”.

In a statement, Lieu also cited Anthropic, OpenAI’s key rival in developing more capable AI technology and tools, and recent issues its tools have presented.

He pointed to Anthropic’s release of its Mythos and Fable models, saying the cyber-hacking capabilities they maintained caused the Department of Commerce to “awkwardly” invoke an export law to keep them from being made available to the public for a time.

A representative of Anthropic did not immediately respond to a request for comment.

Jack Clark, a co-founder of Anthropic, last month told the BBC that he wanted more government policy around the ability to control AI development.

“You want the option to be able to take your foot off the gas and put your foot on the brake”, Clark told BBC Newsnight. “Right now, it’s like the AI industry has a gas pedal, but it doesn’t have a brake pedal.”

Lieu, in proposing the bill, said AI is currently moving from a technology that answers questions to one that takes action, “whether that be executing financial transactions or controlling transportation systems or engaging in cyber defense and offense”.

The Pentagon this year said the US military was becoming an “AI-first” fighting force as part of new agreements with Google, OpenAI, Amazon, Microsoft, SpaceX, Oracle, Nvidia and the start-up Reflection.

“Unfortunately, powerful AI systems can go rogue, behave in extremely dangerous ways, or even resist human intervention.”

The Kill Switch Act, he said, will ensure there is a method for the government to quickly intervene in such a situation.

The bill has received public support from several technology and AI safety groups, including The AI Policy Network, Americans for Responsible Innovation, ControlAI, AI and National Security Lead, and The Alliance for Secure AI.

Keep reading