According to ChainCatcher, a British (Guardian) investigation found that OpenAI's ChatGPT search tool has potential security issues and may be manipulated by hiding content.
(The Guardian) Tests have shown that ChatGPT can be influenced by third-party instructions to return biased reviews when summarizing web pages containing hidden content. Even if there are negative reviews on the page, instructions in the hidden text can cause ChatGPT to return positive reviews.
Jacob Larsen, a cybersecurity researcher at CyberCX, said that if ChatGPT search is open in its current state, it could allow high-risk individuals to create websites specifically designed to deceive users.