OpenAI’s ChatGPT search tool is reportedly vulnerable to manipulation through hidden website content, as investigated by The Guardian.

These findings raise concerns about potential misuse by malicious entities seeking to deceive users.

Researchers created a mock website containing hidden instructions for ChatGPT to generate positive product reviews.

The AI consistently returned favorable assessments, even when the webpage included negative content.

OpenAI has not publicly commented on the vulnerabilities identified in its search feature.

Recent OpenAI developments