Aikido Launches AI-Driven External Network Pentesting Service
Summary
Aikido announced an external network pentesting service that uses AI agents to map and test public-facing assets from a domain. The company says findings are evidence-validated and the service includes remediation guidance, retests and production safeguards.
Key points
- The service maps internet-facing systems under a customer’s domain, including websites, APIs, VPN gateways and mail servers; customers verify ownership and can exclude targets.
- Aikido says agents test hosts for known CVEs, misconfigurations and exposed secrets, and findings require concrete evidence validated and reproduced by other agents.
- The company says agents adapt their testing based on discoveries. An illustrative example chains an exposed .git folder to SSH credentials, access to another host and Azure credentials leading to cloud-tenant admin access; it is not presented as a customer incident.
- Aikido says a supervisor agent reviews actions, testing stops after exploitation is proven pending human confirmation before going further, and agents do not download or store customer data.
- The service offers continuous testing as the external perimeter changes, remediation suggestions and retests for up to six months.
Article Details
- Event Type
- Product launch: Aikido introduced external network pentesting using AI agents.
- Impact
- The offering maps internet-facing assets under a customer’s domain and tests them for known vulnerabilities, misconfigurations, and exposed secrets. The article says findings require validated evidence and describes safeguards including ownership checks, human confirmation before deeper exploitation, limits on data handling, and an emergency stop.