Palo Alto Networks Prisma AIRS adds native runtime security integrations for Claude Enterprise (Anthropic Inference Hooks) and OpenAI Codex Enterprise

Palo Alto Networks' August 2026 Prisma AIRS release notes and vendor social posts (week of Aug 19-21, 2026) confirm two native integrations: Prisma AI Runtime Security plugs into Anthropic's Inference Hooks to inspect/govern Claude prompts pre-inference across Claude, Claude.ai, and Cowork, and Prisma AIRS Runtime API natively integrates with OpenAI Codex Enterprise for real-time DLP/threat prevention on developer prompts without plugins or traffic steering.

This is one of the first commercial third-party security-vendor integrations directly plugging into the frontier labs' own hook/enterprise-admin surfaces (Anthropic Inference Hooks, OpenAI Codex Enterprise admin), giving enterprises inline governance over two major coding/inference assistants without custom proxying.

Security teams running Prisma AIRS who have deployed Claude Enterprise or OpenAI Codex Enterprise should evaluate enabling these integrations for prompt-level DLP and threat prevention.