OrcaRouter Releases Free Firewall and Guardrails for AI Agent Security

Continuum AI has released new security tools for its OrcaRouter platform, introducing Firewall and Guardrails to protect AI agents from prompt injection and related attacks. The announcement was made alongside the AI Threat Report 2026, which warns that prompt injection is becoming the phishing attack of the AI era, according to a press release.

The report states that attackers are increasingly targeting AI systems through emails, documents, and websites that agents process and act upon. OrcaRouter's new tools aim to address these risks by controlling both what agents see and what they can do.

Guardrails monitors prompts and model outputs for prompt injection, jailbreaks, sensitive data leaks, and policy violations. Firewall manages agent permissions, including tool calls, network access, MCP actions, and spending limits. Both components operate inside the OrcaRouter AI gateway and can be enabled without modifying application code.

Firewall and Guardrails are now available at no additional cost for all OrcaRouter users.