Aikido Says Its Code Audit Found More Benchmark Vulnerabilities Than Mythos at Lower Cost
Summary
Aikido says its code audit found 68 of 89 known vulnerabilities in a private benchmark app, versus 60 for Claude Security with Mythos and 58 for Codex Security. Aikido reported a $75 cost, compared with $157 for Claude.
Key points
- Aikido tested three AI code security products against the same private application, which contained 89 known vulnerabilities.
- Aikido reported finding 68 vulnerabilities (76%), compared with 60 (67%) for Claude Security with Mythos and 58 (65%) for Codex Security.
- The vendor says its audit found a critical cross-tenant authorization flaw that could let a user change their organization assignment and access another organization's data.
- Aikido also says it found a timestamp-format mismatch that could leave expired API keys valid.
- Aikido reported a $75 cost at its recommended scan depth, versus $157.06 for the Claude Security run; it says Claude's cost was not shown before scanning.
Article Details
- Event Type
- Comparative benchmark of AI code security products
- Impact
- Aikido reports that, on a private benchmark application with 89 known vulnerabilities, Aikido Code Security Audit found 68 vulnerabilities, compared with 60 for Claude Security with Mythos and 58 for Codex Security. Aikido says its run cost $75, compared with $157.06 for Claude Security and $125 for Codex Security. The benchmark included a critical cross-tenant authorization flaw and an issue that could leave expired API keys valid; these are benchmark findings, not reported production incidents.