When AI Mistook John Galliano for a Basketball Player: Lessons in Information Verification for Vietnamese Sports Journalism

The storm of rumors passes, only verified numbers remain – that is the sentence I remind myself every time I read a sports news piece. Today there are no transfer stories, but there is a notable incident: an automated content analysis system classified an article about the Met Gala – one of the world's biggest fashion events – under the basketball category. Why? The name of the Metropolitan Museum in New York, abbreviated as 'The Met', is exactly the same as... something related to basketball. Specifically, 'Met' might suggest the New York Knicks? Not exactly, but in the database, 'Met' is an important keyword. And the AI made its decision. I laughed, but that smile quickly faded. Because I remembered 2026, at the World Cup in Russia, when I almost believed a football news story from a trusted Twitter account: a Brazilian player named Lucas Almeida was supposedly about to join Everton for £50 million. I had already prepared an entire analysis when a lawyer friend said: 'Have you seen the contract?' – No. In the end, that number was a trick by an investment fund to inflate asset values. I call that 'smart rumor'. And today's AI systems create much dumber rumors. What actually happened? Let's look at the analysis I received from a colleague – an analysis of this classification failure. It points out that the original article was about John Galliano, the designer at Maison Margiela, and about the cancellation of the Met Gala due to an antisemitism controversy. Not a single word about basketball. But the system still labeled it as 'basketball'. Why? Because it saw the words 'Met' and 'Gala' and perhaps linked them to the Detroit Pistons? No, I guess it analyzed context from old articles that had 'Met' in sports headlines. This is the disease of automatic classification: it relies on surface keywords, not semantic understanding. For sports journalists, we constantly face mislabeled information. For example, a story about a 'transfer' in soccer could be mistaken for a 'bank transfer'. But in Vietnam, sports news websites are beginning to use AI to aggregate stories, and this risk is not far off. Let me tell you how I handle rumors. Before writing a transfer analysis, I always ask three questions: (1) Where does this source come from? An official site or an anonymous account? If there's no source, treat it as nonexistent. (2) Does the figure match financial reality? An A-League player is never worth £50 million, no matter how bright the future of football is. (3) Who benefits when this news appears? The agent? The club? The sponsor? Those three layers prevent me from becoming a victim of people with a strategic agenda. An AI misclassification is like an agent lying about a player's value – it serves some agenda. But here, the agenda is the algorithm, not malicious, just lazy. Should we worry? Many people think AI is just a tool and cannot replace creativity. I agree, but if we don't address such errors, one day a sports headline might read: 'Player A suffers a torn ligament after attending the Met Gala' – a meaningless combination. And readers will believe it because they are taught to trust the newspaper. In Vietnam, we have the V-League, we have the AFF Cup, and fans are very passionate. Every transfer window, sports news sites rush to post rumors. I once saw an article about a foreign player rumored to earn $1 million a month, merely because a tabloid reported it. The actual contract was only VND 200 million a month. But because 'rumors' are more attractive than the truth, readers click. That is why we need journalists like me – those who have been burned by rumors and know all too well the limits of sources. But is anyone listening? In the high-speed world of AI, the fastest person is the copy-paster, not the verifier. I end this article with a question: If an AI system can confuse a glamorous fashion event with a basketball game, then where do you trust it to analyze a ball flying into the net? Remember, signatures weigh more than oaths, and an algorithm never signs your article. Let humans do the rest.