Google was forced to urgently disable the AI image generator based on the Nano Banana model, integrated into Google Earth. The tool lasted less than 24 hours — such a rapid shutdown of functionality indicates serious miscalculations in content moderation at the launch stage.

The essence of the innovation was that users could use a text prompt to "draw in" any objects or events on the map. However, within the first hours of operation, the mechanism became a source of outright destructive content. Generated images depicting an explosion crater in Los Angeles and a flooded U.S. Capitol quickly spread across the internet. This is not just a technical glitch, but a direct threat to information security, given the viral potential of such fakes.

The reaction of the professional community is also telling. Open data researcher Henk van Ess discovered during testing that his prompts, including generating scenes with refugees at the Mexican border and images of a nuclear facility in Iran, encountered no blocks whatsoever. This indicates that filtering systems were either absent or configured so superficially that they did not even recognize obviously sensitive topics.

The developers explained the removal of the tool by saying that users were publishing AI-generated images that violated the service's rules. However, such a formulation looks more like an attempt to smooth things over — the problem is not in user behavior, but in the security architecture of the product itself, which was released without proper vulnerability testing.

My assessment: This incident is yet another reminder that the race to integrate generative models into mass-market products often outpaces the development of adequate control mechanisms. For Google, which positions itself as a responsible leader in AI, such a failure is especially reputationally painful. The market and regulators are watching such cases ever more closely, and the rush to release could result not only in technical but also legal consequences.