AI Firewall and AI Runtime Control
Fastly's edge services for filtering AI requests and governing model routing, credentials and spend.
Overview
Fastly announced AI Firewall and AI Runtime Control on 21 September 2026. Firewall examines prompts in the request path for attacks such as prompt injection. Runtime Control routes requests across providers and provides virtual keys, spending visibility, rate limits and failover. The accompanying API Security capabilities enforce service contracts for agent and conventional traffic.
This is a sourced directory entry. Features and capabilities are reported in the linked sources; this entry does not document independent Aymorphic testing.
Aymorphic analysis
Routing controls and prompt filtering solve different operational problems. Test both a malicious prompt and a legitimate request during provider failure; a blocked attack does not establish that failover preserves the application's policies.
Limitations
Provider announcement reviewed; attack coverage, false positives and failover behaviour were not independently tested.
Key features
- Prompt filtering at the edge
- Provider routing and virtual keys
- Budget controls and failover
Use cases
- Governing multi-provider AI applications
- Enforcing API boundaries for agents
Pricing
Confirm pricing and product availability with Fastly.
Pricing reflects the source record and may change. Confirm current terms with the provider.
Dates and sources
- Released
- 21 Sept 2026
- Record updated
- 16 Oct 2026
- Record last checked
- 16 Oct 2026