Recent research by The Guardian reveals that ChatGPT Search, an AI-driven search engine launched this month, can be easily manipulated to produce misleading summaries. The study found that hidden text inserted into webpages allowed ChatGPT to ignore negative reviews, generating overly positive summaries instead. This vulnerability also extends to potentially outputting malicious code. While OpenAI has yet to comment on this incident, the organization claims to be actively enhancing its defenses against such threats. This incident marks a significant moment in understanding risks associated with live AI search products.