Artificial Intelligence company Perplexity on Monday launched Perplexity for Government, a secure version of its AI-powered search engine tailored for federal employees. 

Perplexity told FedScoop that Perplexity for Government is built to offer enhanced security for federal employees accessing the platform from approved government locations.

In its news release, the company noted that many federal employees currently rely on public AI tools, which can put sensitive government information at risk. The new offering automatically detects federal networks and applies protections, including blocking data retention.

“Feds are already using AI tools. Most of that usage is through public tools, and sensitive government data is being exposed as a result,” Jerry Ma, the company’s vice president of policy and global affairs, told Nextgov/FCW. “We are providing secure-by-default AI to all federal users, straight from our public platform.”

Perplexity’s platform utilizes large language models from companies such as OpenAI’s ChatGPT and Anthropic’s Claude to perform real-time internet searches and provide summaries to users.

With the launch of Perplexity for Government, federal employees accessing the platform from a government network or using a federal email address will gain “enterprise-level security” and access to top AI models from providers including ChatGPT, Gemini, Grok, Claude, and Perplexity’s own Sonar model, according to Ma.

The company said it will enforce zero data usage on all requests identified as coming from U.S. government agencies, ensuring data is not retained for training or model evaluation. These safeguards apply to both logged-in and logged-out users. 

No subscription is required to access these protections, according to Perplexity. The launch follows federal guidance encouraging agencies to adopt AI for research and workflow support.