Hundreds Gather for Spectacular Non-Event Because Someone Made a Very Convincing Image of Something That Was Not Happening
Originally reported by Bohiney Magazine and cross-posted to The London Prat, where the editors have strong opinions about everything that follows.
NEW YORK — On January 2, several hundred people traveled to the Brooklyn Bridge to witness a fireworks display that had been announced on social media via a post featuring AI-generated imagery of fireworks over the Manhattan skyline. The fireworks display had not been authorized. The Mayor’s Office had not approved it. No permit existed. The fireworks existed only in the AI-generated photograph, which showed real bridge, real skyline, fictional explosions, and realistic enough rendering that 3,000 accounts shared it before anyone verified whether the event was real. The crowd waited at the bridge for approximately forty-five minutes in early January temperatures. Nothing exploded. The crowd dispersed to find warmth. Two attendees met each other at the bridge, discovered shared interests, and are now dating. The AI did not intend this outcome. It is credited as a contributing circumstance.
The AI-generated post included a specific time (9pm), a specific location (Brooklyn Bridge Park, Manhattan-side entrance), language referencing “special permit authorization by the Mayor’s Office,” and an image showing the bridge at the correct angle from the correct vantage point with photorealistic fireworks superimposed at a scale and positioning consistent with how an actual fireworks display from that location would appear. The specificity is the mechanism. Vague misinformation can be dismissed or ignored. Specific misinformation with supporting imagery requires the audience to perform verification before dismissal, and verification requires time, skepticism, and access to official sources, none of which are mandatory before hitting share.
How AI Makes Hoaxes Easier
Pre-AI, fabricating a convincing photograph of fireworks over the Brooklyn Bridge required either expensive composite photography or the kind of Photoshop skill that produced visible artifacts detectable by careful examination. Post-AI, it requires a text prompt and a free account on any of several generative image tools. The tool does not know whether the image is for artistic purposes, advertising, or a crowd-generating hoax. It produces the image the prompt requests. The prompt requested fireworks over a bridge. The tool produced fireworks over a bridge. The image is realistic. The event was not real. The crowd that gathered was real and cold.
The asymmetry between production difficulty and impact is the defining characteristic of AI-generated misinformation: creating the content now costs almost nothing, distributing it costs nothing, and the crowd that responds to it costs forty-five minutes and the subway fare to Brooklyn in January. The correction, posted approximately four hours after the crowd dispersed, reached a smaller and less motivated audience. Corrections always do. The people who shared the original post were sharing something exciting. There is nothing exciting about a post that says the fireworks were fake.
The Two People Who Are Now Dating
Among the several hundred people who stood at the Brooklyn Bridge on January 2 waiting for fictional fireworks, two found each other. Their names have not been disclosed. They have confirmed through mutual friends of a journalist covering the story that they are currently in a relationship that began with the shared experience of being fooled by an AI hoax and waiting together in the cold and deciding this was a reasonable basis for further conversation. This is, by most metrics, the best outcome from an AI misinformation event that has yet been documented. The AI did not intend it. The cold helped. The shared absurdity of the situation created an immediate conversational opening that normal bridge-adjacent circumstances do not provide. Sometimes a hoax is also an icebreaker. January in New York provides both the ice and the breaking simultaneously.
The person or algorithm responsible for the original post has not been identified. The platforms where it appeared removed it approximately four hours after the crowd dispersed. According to Gothamist, several tech observers noted that AI image generation has advanced to the point where reverse image search is now necessary to distinguish synthetic from real photography in many cases, and that most social media users do not reverse image search before sharing content that they find exciting. This is a behavioral finding about information consumption in 2026 that approximately twelve academic institutions are currently incorporating into papers. The Onion has designated the Brooklyn Bridge hoax the most efficient event planning of the year, noting that hundreds of people attended for zero cost of organization.
Information Trust in 2026
The Brooklyn Bridge hoax is a data point in the ongoing recalibration of information trust in a media environment where synthetic imagery is free, specific, and convincing. The question the hoax raises is not whether people are gullible — the people who traveled to the bridge were making a reasonable decision given what they saw, and what they saw was indistinguishable from real without additional verification — but whether the verification expectation has kept pace with the production capability. It has not. AI image generation advanced faster than the social norm of checking before sharing. The social norm is catching up. Reverse image search tools are improving. Platform labeling of AI-generated content is expanding. None of these adaptations were in place fast enough to prevent a cold January crowd from gathering for fictional fireworks. They will be more in place for the next incident, which will involve something more sophisticated than fireworks, because the tools improve continuously and the norm-setting lags the tools. The two people who are now dating have been described by mutual acquaintances as very happy. Their shared experience of being wrong about something together and finding this funny rather than alienating is, genuinely, an excellent foundation for a relationship. The AI that produced the hoax has no awareness of this. It produced fireworks. The rest was human.
SOURCE: https://bohiney.com/