Censys finds exposed AI infrastructure rose 63% in nine months to over 294,000 IPs, with Langflow and LiteLLM tied to actively exploited flaws.
This story is only covered by news sources that have yet to be evaluated by the independent media monitoring agencies we use to assess the quality and reliability of news outlets on our platform. Learn more here.
More than 294,000 public IP addresses exhibit AI tools and language models, an increase of 63% in nine months. Censys highlights the growth of Langflow and LiteLLM, two platforms with serious vulnerabilities. In the case of LiteLLM, researchers reported targeted attacks against a pre-authentication SQL injection failure.