Skip to main content
AI Security Review is in early access and available on request. Contact Aptible Support to request a review for your organization.

​
What is AI Security Review?

AI Security Review finds exploitable security problems in the applications an organization runs on Aptible. It reads the code and dependencies in docker images built for your Apps, combined with what Aptible knows about how they are deployed and connected. That added context lets the review look for problems a traditional code-only scan or pen-test misses:
  • A flaw in an internet-facing app that an outside attacker can reach directly.
  • A path from a public app into a private service, such as an internal API or database that is not exposed on its own.
  • Chains of the two, where a weakness in one app opens the way to a weakness in another.

​
What the review DOES NOT do

The review reads code and deployment facts. It never touches a running system:
  • It never builds or runs the organization’s code, in any environment.
  • It never sends traffic to running apps, and never queries live endpoints.
It also never examines live secrets; it redacts any secret-shaped files and references in your code, and only examines one-way-hashed secrets in environment variables.

​
How a review works

A review starts when your organization opts in. Aptible chooses which of the organization’s apps are in scope. The review runs in an isolated sandbox on Aptible’s infrastructure. The sandbox contains no credentials, and it has no outbound network access beyond the AI model the review uses. Findings are kept confidential to Aptible’s security team and the organization. Sandboxes are build to meet or exceed best practices for sandbox agent containment. Humans on Aptible’s security team review every report before it is delivered. Each review takes a few business days.

​
What teams receive

Results arrive as an encrypted PDF report. Each finding in it describes:
  • its impact: what an attacker could do and to what;
  • what limits it, such as access an attacker would need first;
  • what to fix.
In early access, Aptible shares the report with users your organization names in your request. Once available in GA, Account Owners will be able to request scans on demand and download the secure report through the dashboard, including reports generated during early access.

​
Request early access

To request a review, contact Aptible Support and ask for early access to AI Security Review. To report a vulnerability in Aptible itself, see the Responsible Disclosure Policy. Aptible AI Security Review is in early release. Features and capabilities are subject to change — contact Aptible Support to provide feedback.

​
Learn more about security on Aptible

Security & Compliance Overview

Learn about the security features included with every Aptible stack

Image and Vulnerability Scanning

Lint Dockerfiles and scan images in CI/CD before deploy