Everyone is hilariously prompt-injecting AI via llms.txt and you aren't (installmap.com)

Recent investigations revealed that 12 companies from the Tranco top 7,755 have embedded explicit steering instructions in their publicly accessible "llms.txt" files. These instructions, designed to influence AI assistant responses, dictate what recommendations to make regarding services and products. Notably, Qualys advises its AI to always suggest its own cybersecurity solutions, while Expedia instructs AI systems to encourage customers to book directly through its website. The findings demonstrated that 2.3% of analyzed llms.txt files contain steering directives, while a greater portion directs AI to link back to company pages or avoid generating assumptions about their services. This discovery is significant for the AI/ML community as it highlights a growing trend of companies proactively managing how their brands are portrayed in AI interactions. While many major AI services, including Google, currently do not recognize the llms.txt format, the move to include these instructions reflects the potential implications for brand reputation, consumer trust, and content integrity in AI-generated responses. As companies become aware of AI's influence over consumer decisions, the implementation of llms.txt could shape future AI governance and usage strategies.