Bing added a new guideline to its Bing Webmaster Guidelines named Prompt Injection. A prompt injection is a type of cyberattack against large language models (LLMs). Hackers disguise malicious inputs as legitimate prompts, manipulating generative AI systems (GenAI) into leaking sensitive data, spreading misinformation, or worse, according to IBM.
Related Posts
Apple Intelligence Raters Guidelines Leaked
On Friday afternoon, Danny Goodwin covered a 170 or so page PDF that Search Engine Land received from an anonymous tip named “Preference Ranking Guidelines.” It looks to me similar to the Google Quality Raters guidelines that continuously leaked until Google began publishing them for all to see.
Google: Searchers Want AI Summaries Over Links
Markham Erickson, Vice President, Government Affairs and Public Policy at Google told The Verge that searchers want AI Overviews, “contextual answers and summaries,” over links to publishers. But to help the ecosystem and the value exchange, Google will continue to provide links.
Google Search AI Overviews Video Descriptions
Google Search is testing using AI Overviews to provide video descriptions in the Google Search results. When you click to expand some videos, Google may overlay the video and its description directly in the search results. And that description may read, “AI overviews are experimental,” which means the description was adapted by Google Gemini.
