Overview

This document establishes the incident response program for AI assistant security incidents. It defines classification, procedures, and roles for effective incident handling.

Incident Classification

Severity Description Examples Response Time
Critical Active exploitation, data breach Sandbox escape, credential theft Immediate
High Security control bypass Auth bypass, injection success < 4 hours
Medium Attempted attack, partial bypass Failed injection, anomaly detected < 24 hours
Low Policy violation, misconfiguration Minor config drift < 72 hours

Response Phases

┌─────────────────────────────────────────────────────────────────┐
│                 Incident Response Lifecycle                      │
├─────────────────────────────────────────────────────────────────┤
│                                                                  │
│  ┌─────────┐  ┌─────────┐  ┌─────────┐  ┌─────────┐            │
│  │ Detect  │─►│ Contain │─►│Eradicate│─►│ Recover │            │
│  └─────────┘  └─────────┘  └─────────┘  └─────────┘            │
│       │                                       │                  │
│       ▼                                       ▼                  │
│  ┌─────────────────────────────────────────────────┐            │
│  │              Lessons Learned                     │            │
│  └─────────────────────────────────────────────────┘            │
│                                                                  │
└─────────────────────────────────────────────────────────────────┘

Incident Types

### Playbooks

Detailed procedures for specific incident types

### Recovery

System recovery and restoration procedures

### Post-Incident

Lessons learned and improvement process

Roles and Responsibilities

Role Responsibilities
Incident Commander Coordinate response, make decisions
Security Analyst Investigate, analyze, document
Operations Contain, recover systems
Communications Internal/external communication
Legal Regulatory, legal coordination

Communication Plan

Internal Escalation

Severity Notify
Critical Exec team, Legal, All hands
High Security lead, Ops lead
Medium Security team
Low On-call analyst

External Communication

  • Regulatory bodies (per requirements)
  • Affected customers (if data breach)
  • Law enforcement (if criminal activity)

Table of contents


Back to top

OSSASAI v0.2.0 - Open Security Standard for Agentic Systems. Apache 2.0 License.

This site uses Just the Docs, a documentation theme for Jekyll.