How Malwarebytes Secures Its Privileged Antivirus Software

Summary
Malwarebytes describes securing its privileged Windows antivirus with per-call authorization, threat modeling, code analysis, AI-assisted review, fuzzing, structured remediation, signed updates, and regression testing. It also discusses AI’s role in vulnerability खोज.
Key points
- Malwarebytes says its Windows product separates a user-level interface from a SYSTEM-level service; the service must authenticate and authorize each privileged request, including requests to change exclusions or protection settings.
- The company describes threat modeling, least privilege, minimal interfaces, secure coding checks, static and dynamic analysis, fuzzing, and continuous scans for exposed secrets and vulnerable dependencies.
- Malwarebytes says it uses an internal AI scanner and an AI agent to review pull requests for security issues, with engineers validating findings and proposed fixes.
- The company says findings are reproduced, triaged by severity, assigned for remediation, and followed by regression tests and updates to fuzzing and scanner rules; fixes are delivered through signed updates.
- Google reported that its Big Sleep AI agent discovered SQLite flaw CVE-2025-6965, which Google said stopped attackers from exploiting it. Mozilla confirmed 22 security-sensitive bugs found through work with Anthropic, 14 rated high severity.
- OpenAI reported that an autonomous pipeline using an unreleased model built working root exploits for 948 unique Linux kernel local privilege escalation vulnerabilities in one scan. HackerOne reported a growing backlog of valid, unresolved findings.
Article Details
- Topic
- Malwarebytes' security engineering practices and the impact of AI on vulnerability discovery and remediation
