Google Earth: Twenty Years to Build Trust. Twenty-Four Hours to Put It at Risk

Google Earth has existed since 2005. In twenty years it has become the visual reference source for journalists, researchers, investigators and courts. Even the United Nations uses this tool, which has earned a fundamental evidentiary role in global dynamics.

After all, a satellite image is credible because it cannot be invented. And over the years this is what made Google Earth different from any other platform.

In twenty-four hours, however, Google put all of this at risk by integrating Nano Banana 2 into Google Earth.

Why did it do it? To let users indulge their imagination, allowing them to select any point on the planet and modify it. A creative outlet that should have made it possible to visualize the Colosseum at the height of its splendour, or Pompeii during the eruption. And yet, that is not what happened.

Within a few hours, around the world appeared: King Kong, nuclear power plants in Iran; Mexican refugees along the US border; nuclear craters in Gaza. Airplanes (obviously, against New York skyscrapers).

No sensitive prompt had been rejected. A clear governance error, which forced Google to withdraw the feature in less than twenty-four hours, putting its reputation at risk.

The underlying error

The decision that produced this result is a clear category error.

Whoever approved the integration of Nano Banana 2 into Google Earth treated Google Earth as a creative platform: which Google Earth is not. Adding an AI image generator to a creative platform would have been a feature extension. Adding an AI image generator to an evidentiary standard, instead, is an attack on its primary function. And therefore on its entire architecture of value.

Is it possible that Google did not know the real function of its own tool? Or did it really think people would not use the feature in ways like these?

The defense that does not defend

Google’s initial response to the criticism was a watermark: every generated image contained SynthID, the invisible marker for the company’s AI products. A useless response, because during a conflict any false satellite image can still destabilize. In the rush of the moment, nobody checks the watermark (and we had already seen this at the beginning of 2026 in the war between the United States, Israel and Iran). Nano Banana 2 would have made that process trivial.

The watermark depends on “correct” use. But what about someone who wants to spread disinformation? Using it as a security argument, therefore, shows how little the problem was understood. And Google, clearly, did not understand much of it.

No testing, no protection

It is clear that nobody asked, before launch: “What would happen if a user generated a nuclear power plant in Iran?”

Yet that is what testing is for: particularly in AI, where the sophistication of the tool necessarily requires deeper reflection.

Eliot Higgins, founder of Bellingcat, wrote that “Google launched a tool that appears not to have passed the most basic adversarial tests before going into production. Those tests were then carried out by strangers, for free, at Google’s expense.”

Governance of an AI service cannot be a phase that comes after launch. It has to begin before launch, before any damage to trust.

King Kong climbing Taipei 101. A playful prompt that certainly will not fool anyone. The same cannot be said, however, of missiles in Cuba or nuclear plants in Tehran.

The lesson for any organization with a trust asset

Google will get through it. It is such a large, deeply rooted organization that it can withstand even a crisis such as Nano Banana 2 on Google Earth, withdrawn within twenty-four hours.

Not many other companies, however, could withstand the same phenomenon.

We know well the pressure that today pushes organizations to integrate AI capabilities into every existing product and service (after all, we discussed it in the article on AI Washing). But one should not give in to the temptation to do so regardless of the product category. And at the expense of service quality.

The governance question to ask before any AI integration is not only “what can this tool do?”, but: “Which property of the product makes what we already do credible?”

If the integration preserves value, then it is right to proceed. If it erodes it, or there is a possibility that it will erode it, it should undoubtedly be avoided.

Google can withstand similar impacts. For smaller organizations, that is not guaranteed.