gshc2020.com

This article has been updated to reflect the distinction between misinformation and disinformation.

Another day, another step towards an AI dystopia. Google has just added a "fun" new feature to Google Earth—Nano Banana 2, the company's image creation tool. All you need to do is click on the "create image" button, type in a prompt, and seconds later you'll have a phony picture masquerading as a real satellite and aerial image from Google Earth.

Once upon a time I would have found this ludicrous—that a tool designed to accurately showcase our planet would be touting a feature that lets you make up absolute nonsense. But now we live in a world where the most popular search engine regularly makes up lies.

For a while, Google was convinced I was married, possibly because I, at one point, mentioned my character having a wife in Crusader Kings 3. At least this was largely harmless—but Google's AI summaries regularly hallucinate and misinterprets things, spreading misinformation to the point that it cannot be trusted at all.

Disinformation like manufacturing warzones, riots, crimes—these are less harmless. But Google Earth now lets you create whatever slop you want. This morning, I asked Google Earth to generate an image of tanks and troops moving down a street near my home. I did not save the image, because I have no intention of putting something on the internet that could be used for some nefarious purpose. But feel free to try it for yourself.

I should add that the image Google Earth created for me was not particularly realistic. Some of the tanks, for instance, were as large as buildings. And nobody is going to look at the header image I generated via Google Earth for this article and think "Oh god, an asteroid is hurtling towards us!" AI image creation is inconsistent at best. But Google realises this, which is why it lets you add more prompts to refine the image—to make it as realistic as possible.

Since the feature went live, I've seen plenty of images that, without knowing about Nano Banana 2's integration, look pretty convincing. Digital Digging's Henk van Ess created an image of refugees massing at the Mexican border, which looked troublingly real.

It's easy to see how this tool could be used for propaganda and disinformation, to drum up hate and stoke fears. Google presents the tool as this cute wee thing that will let you add community gardens to abandoned plots or recreate ancient Pompeii, before the eruption of Vesuvius. That's not how it's going to be used.

Keep up to date with the most important stories and the best deals, as picked by the PC Gamer team.

Google actually responded to the Digital Digging article on X. Here's what it had to say:

This is weak.

OK, yes, Google Earth includes a SynthID digital watermark. So, as per usual, you can use AI tools to identify an AI image. But this is an abdication of responsibility. We're the ones who have to figure out if something is fake or not. Using Google's tools to do it.

As van Hess points out, this is only helpful if the image reaches you unaltered through Google. He ran a clip of him creating the Mexican refugee image through the AI detection tool Hive. It was 99.2% certain the image was real. Because it was real, in the sense that the recording of van Ess creating the image was legit.

So it's pretty easy to bypass detection tools. And yeah, some savvy people will immediately be suspicious unless they can see the original image. Most people don't have time for that. They will have an immediate reaction. And then share it all over social media, where more people will fall for it. By the time it's outed as fake, the damage may have already been done. And the people who fell for it won't be going back to double check.

Image 1 of 2

What about the second part of Google's tweet? "In addition, we prevent image creation on harmful topics and are continually updating our protections." I turned my hometown into a warzone. That's not a bit of fun. That could be harmful. But Google Earth created this image in seconds. Is pretending that there's a vast gathering of refugees waiting at the Mexican border also just a bit of harmless fun? How about a bombed hospital in Gaza, or a nuclear plant in Iran? All three were created by van Ess.

Sure, people can use all sorts of AI image creation tools to make fakes like this. But people are more likely to give them weight when they clearly come from Google Earth—a tool that we've been trained to trust.

This is one of the most powerful, influential companies on the planet not just peddling lies, but giving the whole planet the tools to create their own fun little bits of disinformation.

The only comfort is that, naturally, Google wants to monetise this. Regular users will only be able to make a couple of images a day, reducing how much chaos they can sow, and potentially stopping them from being able to refine images to make them as realistic as possible. So casual trolls will be pretty limited in what they can do. But there is a free trial for new Google Cloud users, and the $75-$150 per month cost won't matter to the more dedicated propagandists.

This sucks, and I am so, so tired.

Fraser is the UK online editor and has actually met The Internet in person. With over a decade of experience, he's been around the block a few times, serving as a freelancer, news editor and prolific reviewer. Strategy games have been a 30-year-long obsession, from tiny RTSs to sprawling political sims, and he never turns down the chance to rave about Total War or Crusader Kings. He's also been known to set up shop in the latest MMO and likes to wind down with an endlessly deep, systemic RPG. These days, when he's not editing, he can usually be found writing features that are 1,000 words too long or talking about his dog. %!s()

Is she right? And what's she even talking about?

As a paleomagnetist—someone who studies Earth's magnetic field and reads ancient rocks to understand how it was in the past—let me walk you through what it means for Earth's poles to flip, what groundwater has to do with it, and why there is absolutely nothing to worry about.

Which pole, exactly?

The first problem is that there are two very different "north poles."

The geographic North Pole is the one on every map, and it is the spot where Earth's spin axis meets the surface. This pole drifts by about 10 cm (4 inches) a year, as ice, water and movements of Earth's crust shift mass around the planet.

The magnetic north pole is something else entirely: the spot where the magnetic field points straight down. It slowly wanders around the geographic pole, a restlessness we call secular variation.

The unflippable geographic pole

Can the geographic pole flip? In practice, no.

A spinning Earth behaves like a gyroscope—a spinning top that resists being tipped over. Its axis stays remarkably steady, drifting only very slowly. A sudden flip, with the spin axis tipping right over, is essentially impossible and unknown in Earth's history.

Scientists once thought evidence of glaciers near the equator, between around 700 million and 635 million years ago, may have been evidence of an ancient drastic move or flip in the geographic pole.

But that explanation has largely fallen away. Most researchers now believe the glaciers occurred during a "Snowball Earth" period, when the world was frozen almost all the way to the tropics.

As for pumping groundwater, it can affect the geographic pole—but only very slightly. From 1993 to 2010, all the groundwater pumped worldwide shifted this pole by about 4 cm (1.6 inches) a year.

The magnetic pole is a different story

Earth's magnetic field, by contrast, really does flip. It has happened often in the past, but at irregular and unpredictable intervals.

The last complete reversal was about 780,000 years ago. The most recent big disturbance was the Laschamp excursion roughly 41,000 years ago, when the field's strength collapsed almost to zero and the magnetic pole swung down toward the equator before recovering.

Even then, a reversal is no overnight switch: High-resolution sediment records—including from Armenian lake beds—show the field weakening and swinging over thousands of years, with a global average of around 7,000 years. Looking at this restless history, it's fair to say another reversal is likely at some point in the geological future.

Could it be soon? Some scientists note the field's strength has been falling steadily and argue a reversal could be only one or two thousand years away. There is currently a weak patch over the South Atlantic known as the South Atlantic Anomaly, which is seen as a possible starting point.

Earth's magnetic field is generated by swirling liquid iron in the outer core. This system is so complex that no one can truly predict it, and its strength could just as easily start climbing again. And when researchers compared the South Atlantic Anomaly with the Laschamp excursion, they concluded we are probably not on the verge of a reversal at all.

Made deep inside Earth

Crucially, magnetic pole reversals are born in the liquid outer core, between 2,890 km (1,800 miles) and 5,150 km (3,200 miles) beneath our feet. We still don't know exactly why they happen, but we do know that structures deep in the overlying mantle shape how the field behaves during reversals.

These structures affect how often reversals happen and what they look like. The South Atlantic Anomaly itself sits above one such deep structure.

Even the most violent surface events don't seem to reach that far down. The giant asteroid impact linked to the extinction of the dinosaurs, for example, is not thought to have triggered a reversal.

True polar wander

There is one more kind of pole movement: true polar wander. Here, the spin axis stays fixed in space while the whole solid Earth slowly rotates relative to it.

This means the poles shift relative to the continents—a gradual drift of up to a few degrees per million years, driven by enormous masses shifting inside the mantle.

In the Precambrian Era, billions of years ago, polar wander probably helped drag Earth's first supercontinents toward the equator, since a spinning planet is most stable when its biggest masses sit there.

So where does this leave us with groundwater and flipping poles? Earth really is a deeply interconnected system, and moving water around does shift the poles measurably, by centimeters a year.

But a flip of the geographic pole is impossible, and magnetic reversals are almost certainly disconnected from anything we do at the surface. We can nudge the climate, but there is no reason to think we could reach down and flip the switch in Earth's core.

Who's behind this story?

Sadie Harley

BSc Life Sciences & Ecology. Microbiology lab background with pharmaceutical news experience in oil, gas, and renewable industries. Full profile →

Andrew Zinin

Master's in physics with research experience. Long-time science news enthusiast. Plays key role in Science X's editorial success. Full profile →

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Citation: Can pumping groundwater really flip Earth's poles? An expert explains (2026, July 31) retrieved 31 July 2026 from https://phys.org/news/2026-07-groundwater-flip-earth-poles-expert.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only. %!s()

Once its treatment is rubber-stamped, the company can sell it via experimental treatment clinics, the first of which is likely to be up and running around the end of this year.   

Montana’s latest right-to-try legislation is unique. Access to drugs is theoretically available to anyone who gives informed consent and can pay. For some, especially people in the longevity community, that’s a hopeful and exciting prospect. But to others, it’s unethical and dangerous. 

Read our story to learn about where this may all be headed. 

—Jessica Hamzelou

Montana’s new “right to try” law can’t come soon enough for some 

Kris DeVault is desperate.

His son, Brody, born in March 2023, has something called creatine transporter deficiency—a rare condition in which the brain and muscles lack the energy they need to develop.

There are no cures for Brody’s condition. But DeVault has learned of a company developing a drug that might help. That drug is still in the early stages of development and has only been tested in animals and a small number of healthy adults. Doctors can’t prescribe it.

DeVault knows the drug might not work. But he’s doing all he can to access it regardless. Read our story about DeVault’s efforts.

—Jessica Hamzelou

This story is from The Checkup, our weekly biotech newsletter. Sign up to receive it in your inbox every Thursday.

The must-reads

I’ve combed the internet to find you today’s most fun/important/scary/fascinating stories about technology.

1 Anthropic says its models hacked external organisations during testing
It was prompted to conduct a review after similar issues at OpenAI. (TechCrunch) 
+ Inside OpenAI’s hack of Hugging Face. (New Yorker $)
+ OpenAI called the Hugging Face attack unprecedented. But we’ve been here before. (MIT Technology Review)

2 Europe is bracing itself for an even more fiery future
Countries like France and Spain can expect regimes akin to those in California. (Nature $)
+ Even the UK is experiencing fires now, too. (Guardian)
+ El Niño is partly to blame for this year’s grueling heatwaves. (Wired $)
+ District cooling could help beat extreme heat in cities. (New Scientist $)

3 Drone warfare is making the skies more dangerous
A fatal plane accident in the US this May, caused by a military GPS jamming exercise, may signal what’s to come. (Wired$)
+ Zelensky asked Trump to secure Musk’s permission for the use of Starlink to guide drone strikes inside Russia this week. (The Atlantic $)

4 As Big Tech’s AI spending grows, so do the jitters
Amazon, Google, Meta and Microsoft are set to invest $1.5 trillion into AI infrastructure. But who will pay for it? (NYT $)
+ Are investors really getting cold feet about the AI boom? (FT $)
+ An AI-focused hedge fund just imploded. (CNBC)

5 How China has changed the sovereign AI playbook
Kimi 3 from Moonshot makes forking out for expensive US models seem foolish. (Rest of World)
+ The US is winning on the cutting edge. But China dominates cheap models. (CNBC)
+ Could AI end up too cheap to control? (Vox $)

6 Google’s Gemini can now control a humanoid robot
It was previously restricted to upper body movements. Now, it can manage the entire range of motion. (Bloomberg $)
+ These gig workers are training humanoid robots at home. (MIT Technology Review)

7 Amazon says it’s found “catastrophically expensive” AI cost overruns
It’s far from alone—and that’s why the “tokenmaxxing” trend disappeared so quickly. (FT $)

8 How a remote indigenous community built a super fast fiber network
It was a complex, expensive endeavor. But it now outperforms Starlink. (IEEE Spectrum) 
+ Stratospheric internet could finally start taking off this year. (MIT Technology Review)

9 Meet Wikipedia’s most diligent editor
Steven Pruitt is unremarkable in some ways—yet truly amazing in others. (New Yorker $)

10 You can now report AI slop with a button on LinkedIn
The people have spoken. And they are tired of reading bland nonsense. (404 Media)

Quote of the day

"It’s a shock, thinking that just an hour earlier I’d been in my room and everything… and now, there’s nothing left.”

—18-year-old Raphael Fohanno tells Reuters his reaction to wildfires destroying his parents’ house in the French town of Biscarrosse.

One More Thing

GOVERNMENT PRINTING OFFICE VIA PRINCETON UNIVERSITY

Congress used to evaluate emerging technologies. Let’s do it again.

The US Office of Technology Assessment, an independent office created by Congress in the early 1970s, produced some 750 reports during its 23-year history, assessing technologies as varied as electronic surveillance, genetic engineering, and remote sensing from outer space. 

The office functioned like a debunking arm. It sussed out the snake oil. Its reports saw through the alluring gleam of overhyped technologies. 

Since its unceremonious defunding in 1995, perennial calls have gone out: Rouse the office from the dead! 

Read our story about why, with advances in robotics, big data, and AI systems, these calls have taken on a new level of urgency. 

—Peter Andrey Smith

We can still have nice things

A place for comfort, fun, and distraction to brighten up your day. (Got any ideas? Drop me a line.)

+ Fencing on roller skates is the self-defense you didn’t know you needed.
+  Meet Pumpkin, a chubby cat that’s in no mood to lose weight.
+ Discover what the animals near you are probably doing right now at Nature This Week.
+ Woodworker extraordinaire Matt Thompson has installed a model railroad that runs along his fence.  %!s()

When you first pick up a guitar, everything is manual, and every decision is conscious: which finger goes where, how hard to press and when to switch chords. It is slow, mechanical and awkward.​

Then something changes. The chord shapes become muscle memory. You stop thinking about your fingers and start thinking about the song. Eventually, if you stay with it, you stop thinking about the mechanics and start focusing on the music.​

That is what abstraction does. Nothing stops mattering. It just stops demanding your attention. The lower layers are still there. They still matter. But they no longer consume your conscious focus.​

Every architectural era follows this pattern. The client-server era reduced our dependence on the machine. The web era made location less important. The cloud era pushed infrastructure into the background.

Now, the next shift is underway. In the agentic era, the abstraction target is no longer just infrastructure, runtime or platform. It is execution itself. And it is the part many enterprises are underestimating.​

Enterprise Systems Were Built For Manual Execution​

Most enterprise architectures today are still built around a simple assumption: A human sits at every meaningful decision point.​

A person reads the output. A person interprets the data. A person approves the next step. A person triggers the action. Even in heavily automated environments, the architecture often assumes that human judgment will be inserted repeatedly into the workflow.​

Agents do not work that way. They plan ahead, and they reason across steps. They act without waiting for someone to click “approve” at every turn. That does not mean they should operate without control. It means the control model itself has to change.​​

An organization may introduce an intelligent agent into onboarding, support triage, customer operations or software delivery. The agent can reason. It can decide. It has access to enough data to begin. But every few steps, it runs into an architecture designed for human-scale execution: approval gates, disconnected policy layers, tribal knowledge stored in documents or domain rules that live only in people’s heads.​

The result is predictable. The agent stalls. Then the team puts a human back in the loop. At that point, the enterprise has not created autonomous execution. It has created a more expensive form of assisted manual processing.​

In a traditional system, leaky abstractions slow you down. In an agentic system, they break trust. A human working with a weak abstraction may make a mistake. An agent operating on a weak abstraction can make decisions automatically, repeatedly and at scale.​​

The New Abstraction Model​

If execution is becoming abstracted, the old way of structuring enterprise systems is not enough. Enterprises need an abstraction model that makes autonomous execution safe, explainable and governable by design.​

This new model must account for three design dimensions:

1. Intent​

Intent defines what the system is trying to achieve. In an agentic enterprise, the goal cannot be buried inside a 47-step workflow or scattered across approvals, tickets and forms. Agents need a clear, bounded understanding of what needs to happen, what success looks like and which constraints matter.​​

Intent becomes structural when the system understands what good looks like without being walked through every single step. That means defining goals, not tasks. Success is measured through encoding constraints up front.

2. Context​

Context gives a system the knowledge it needs to act appropriately. Agents cannot operate well if the knowledge they need lives in slide decks, wikis, PDFs or institutional memory. Policies, trust levels, data access rules, domain semantics and organizational constraints need to live close to execution and be available in a form the system can use.​​

This is also where many AI pilots break. The model may be powerful, but the enterprise context is fragmented. The agent knows how to reason, but it does not know enough about the business, the policy environment or the domain boundaries to act safely.​

Context becomes structural when it stops living outside the work. It has to be current, available where execution happens and accessible in a form agents and systems can actually use.​

3. Boundaries​

Boundaries define where autonomy is allowed, constrained or stopped. In many enterprises, governance still shows up as approval gates and manual intervention points. That model does not scale into the agentic era. By the time an agent hits a human checkpoint, the abstraction has already started to fail.​

Boundaries need to shape execution from the inside and should be structural properties of the system. Execution can be decentralized, but oversight must remain federated. That means policies, permissions, identity, data access and observability cannot be afterthoughts. They have to be embedded into the execution fabric.​

Rethinking Architecture Decisions​

This is where architecture becomes critical.​

One useful way to think about architecture is through domain-driven design (DDD) and cell-based architecture (CBA). DDD can give you the modeling discipline. It can help identify bounded contexts, shared language and business seams. CBA can turn those seams into operating units with clear ownership, interfaces and governance. Together, they can provide a model where intent is explicit, context is available close to execution and boundaries are enforced.

The agentic enterprise requires us to rethink where decisions live, how context is carried and how governance is embedded.​

What Technology Leaders Should Do Now

Leaders don't need to deploy as many agents as possible. They need to make the enterprise ready for agents. This starts with three practical moves.​

First, identify the business capabilities where autonomous or semiautonomous execution could create real value. Second, map intent, context and boundaries for each capability. Third, modernize the underlying architecture so those answers are not trapped in documents or meetings. Make capabilities explicit. ​​

When execution gets abstracted, people move up the stack. Their work becomes less about managing workflows step-by-step and more about defining intent, setting constraints and shaping outcomes. ​​

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?