Cloudflare extended its secure access service edge (SASE) and zero-trust services to help organizations measure and manage the use and integration of generative artificial intelligence (AI) tools. The extension aims to prevent data loss and open the doors for safe AI technology adoption.

Generative AI tools and large language models (LLM) such as OpenAI’s GPT and Google’s Bard can be used for a variety of tasks, but “every input fed into a prompt becomes a piece of data leaving your organization’s control,” Cloudflare noted in a blog post.

"AI holds incredible promise, but without proper guardrails it can create significant risks for businesses,” Cloudflare co-Founder and CEO Matthew Prince said in a statement. “It is far too easy, by default, to upload sensitive internal or customer data to AI tools. Once the data is used for training AI, it is virtually impossible to get it out."

Several major companies including Samsung, Amazon and JPMorgan already restrict employees’ use of those tools. A recent KPMG report on generative AI also showed cybersecurity (81%) and data privacy (78%) are among the top concerns for implementing AI among U.S. executives.

However, Cloudflare argues organizations shouldn’t have to block the use of generative AI tools, and SASE capabilities with zero-trust security controls can offer the ability to use these tools safely.

Cloudflare's One for AI platform allows the vendor’s SASE customers to apply guardrails by reviewing or even blocking the usage of generative AI tools and managing the integration. Specifically, Cloudflare said it:

• Allows IT teams to deploy a secure web gateway (SWG) feature — Cloudflare Gateway in this case — to passively observe and track AI usage among employees, which can offer context for budgeting and enterprise licensing plans.
• Includes service tokens that enable administrators to maintain a record of API requests. These tokens provide control over which specific services can access AI training data and the ability to revoke tokens when building ChatGPT plugins for internal and external use.
• Includes Cloudflare Tunnel to provide an encrypted, outbound-only connection to the vendor’s network. This scrutinizes every request against access rules set for services under Cloudflare One or those in which teams are ready to allow an AI service to connect to their infrastructure.
• The zero-trust access control can offer the ability to enforce authentication on each request made to the data the organization decides to provide to these tools.
• The data loss prevention (DLP) service can provide basic preconfigured options to check for data that looks like social security numbers or credit card numbers. Additionally, it enables custom scans for patterns based on a specific team's data configurations.

Cloudflare plans to expand its cloud access security broker (CASB) service to scan AI tools to detect misconfigurations and misuse.

The vendor also unveiled several new AI-powered features including new ChatGPT plugins, an experimental AI assistant and an AI-application prediction integration.

The Cloudflare Radar plugin is designed to allow users to talk to ChatGPT about real-time internet patterns powered by Cloudflare Radar, which is a hub that showcases global internet traffic, attack, and technology trends and insights.

And the Cloudflare Docs plugin is designed to enable developers to use ChatGPT to help write and build Cloudflare applications with up-to-date information from its documentation.