The exploding supply of misinformation is not the main problem
Google briefly made fake satellite images easier to produce, which is not great, but the economics of bad information still begins with demand.
For roughly a day and a half last week, Google Earth contained an image generator.
A user could select a real place, type a prompt and ask Google’s Nano Banana 2 model to produce a photorealistic image grounded in the platform’s satellite, aerial and 3D imagery. People immediately started generating scenes of bombings and humanitarian crises that had not happened. Google paused the feature after, in its words, “people [began] sharing screenshots of generated imagery that appear to violate our policies”, and said it would work on stronger safeguards. Some product risks are apparently impossible to anticipate, even for one of the world’s largest technology companies, with almost unlimited engineering, policy and research resources. If you cannot sympathise with Google being surprised that its fake-satellite-image generator produced fake satellite images, you really do not have a heart.
Some people took all of this as the most recent proof that our shared reality is shattering. And while I agree it was a dumb product decision, I am less convinced that it changed the economics of misinformation as much as the response suggested.
Media Finance Monitor is where newsroom leaders, funders, and media entrepreneurs come to understand how money moves through the information ecosystem. We decode funding, business models, policy, and platform shifts and turn them into strategies you can actually use.
Paid subscribers get premium editions, full access to the archive and all our research reports, for about the cost of a pistachio éclair and a good espresso a month.
For deeper engagement, our Partnership Program offers one-on-one strategic advisory, curated briefings, access to our internal databases and invite-only events.
The underlying question is whether we think misinformation is primarily a supply-side or a demand-side problem.
If it is a supply problem, Google has plainly made things worse. Manipulating satellite images was already possible, but it required some skill, software and time, and generative AI now reduces all three. Under normal economic conditions, lowering the cost of producing something will usually lead to more of it being produced.
The same has already happened across large swathes of the information ecosystem. My most recent encounter has been on YouTube, where fake movie trailers, synthetic explainer videos, imaginary celebrity news and rankings of the ten most terrifying medieval kings now proliferate endlessly. (My feed may contain a higher volume of medieval elites than yours, but trust me: the wider phenomenon is real.)
YouTube says mass-produced, repetitive and generic AI content is not eligible for monetisation while synthetic content can earn money when it contains sufficient originality and value. This is a sensible distinction and an enforcement nightmare. There is a huge grey area with material that is not quite fraudulent, not quite useful and engaging just enough to produce some revenue.
The unit economics are attractive. Generative tools reduce the fixed cost of production, platforms provide global distribution, and even a small amount of engagement can be converted into some advertising income. Most individual experiments will fail, but when production is nearly free, a producer can run thousands of experiments.
This is the supply-side effect of AI that matters. Still, supply is only valuable when it meets demand.