The FBI says a 35-year-old man named Love ran a romance-investment scam by posing as an injured San Francisco 49ers receiver, and that when his victims tried to check the story, search engines and AI systems sometimes told them it was true. Court filings describe women who grew suspicious, looked him up online, and were informed that he really did play professional football. Love had manufactured that answer himself, by posting the claim across Instagram, TikTok and LinkedIn in enough volume that the machines reading the internet treated it as established.
The scheme itself is ordinary. According to the FBI, Love presented himself in different versions as either a 49ers player or a wealthy Swiss property developer, while Chan played the part of his financial adviser. They met women through dating apps, steered them into fictitious investments and kept the money. Investigators say some victims were talked into taking out loans to fund them.
What is not ordinary is the corroboration layer. Love posted constantly as an active but injured 49ers wide receiver, and the filings say the sheer quantity of that material gave the scammers unexpected help from search and AI products. The verification step a careful person performs — type the name, see what comes back — returned the lie.
The LA Times reports that Love did not simply benefit from this; he used it. In a since-deleted Instagram post about his "road to the NFL," he published screenshots of a Google AI Overviews answer. The feature stated that Daejon Love was a San Francisco 49ers receiver and that information about his position, his contract and his correspondence with teammates was available online. It cited his connection to the team through social posts and articles. Showing those screenshots, per the LA Times, Love pointed at Google and stressed that these were not his words but the system's own conclusion.
That last move is the whole story, and it is not the familiar complaint about AI hallucinations. The model being wrong is not the interesting part; models are wrong every day and mostly nothing happens. The interesting part is that Love created the inputs, harvested the output, and then presented that output to his targets as independent confirmation from a party with no stake in the outcome. Search engine optimisation aimed at a ranking, which a reader still had to interpret. This aimed at a sentence, delivered in Google's voice, to a woman deciding whether to wire money.
Look closely at what AI Overviews actually asserted and it goes further than repeating a false claim. It said documentation existed — a contract, messages with teammates — that a reader could go and find. The system did not just inherit a lie from its sources; it vouched for evidence that was never there, in the confident register the product uses for everything. There is no hedge in an AI Overview. That is the design.
Chatbots are easy to push toward stating particular products or claims as fact, and an industry has already formed around that property, the natural successor to SEO. People have been poisoning the record for years with fabricated social posts and Wikipedia edits. The difference is that a human researcher weighing sources of unknown quality writes "allegedly"; a generated summary collapses the same material into a declarative line and hands it to someone who asked a yes-or-no question.
Nothing in the filings or the reporting says whether Google corrected the answer, whether variants of it still surface, or how many of the victims made a decision because of it. That absence is the part worth sitting with. The scam has a defendant, a docket and an investigating agency. The machine that supplied its references has no role in the case at all.