Microsoft has introduced a new guideline in its Bing Webmaster Guidelines called “prompt injection” to address the abuse and attack of language models by websites. Prompt injection is a security vulnerability that affects AI and machine learning models, particularly large language models (LLMs), by manipulating the prompt to trick the model into following unintended instructions. This can lead to the generation of fake news stories based on misleading information. The guideline warns against adding content on webpages that attempt prompt injection attacks, as it can result in demotion or delisting from Bing search results.
The new guideline has been added to the Bing Webmaster Guidelines and any webpage found using prompt injection techniques may face consequences such as demotion or removal from Bing search results. This move aims to protect the integrity of search results and prevent the spread of misinformation through manipulated language models. Websites are advised to adhere to this guideline to maintain their visibility on Bing’s search engine.
Source link
Source link: https://searchengineland.com/prompt-injection-added-to-bing-webmaster-guidelines-443788