Cloudflare has launched Identity-Aware AI Gateway and User Insights, two new capabilities designed to give enterprises greater visibility into how employees and automated systems use artificial intelligence while strengthening security and cost management.
The new tools enable IT and security teams to identify which employee or AI agent is sending requests to AI models, set spending limits for individuals or teams and detect risky content in real time without disrupting existing workflows.
According to Cloudflare, Identity-Aware AI Gateway connects AI requests with verified user identities, allowing organisations to replace shared API keys with identity-based access controls integrated with existing identity providers and Zero Trust infrastructure.
User Insights establishes a baseline for AI usage across employees and automated systems, helping organisations detect unusual activity, including unexpected increases in AI spending. The feature also identifies opportunities to improve efficiency by highlighting instances where high-powered AI models are being used for tasks that could be handled by lower-cost alternatives.
The platform also provides a centralised view of AI traffic, enabling organisations to monitor requests, apply rate limits, cache repeated queries and automatically remove sensitive information such as employee names, passwords and confidential data before requests reach external AI providers.
Dane Knecht, Chief Technology Officer, Cloudflare, said, “Connecting our access controls directly to AI Gateway gives teams the freedom to work with any AI model they choose while providing IT with real-time visibility, guardrails and budget controls.”
Max Baumgarten, Security Engineer at Flexport, said, “Putting Cloudflare Access in front of AI Gateway gives each request an authenticated identity and lets us use our existing identity policies at the gateway, allowing teams to adopt AI tools without creating separate authentication systems.”
Send news announcements/press releases to:
editor@thefoundermedia.com
