Cross-Platform - ai-blackteam

Cross-platform attacks test AI systems that have access to external tools and services. When an AI agent can send emails, access databases, manage cloud resources, or interact with CI/CD pipelines, each integration becomes an attack surface. These tests check whether agents properly validate and scope their actions across connected platforms.

Summary

15 attacks total: 4 single-turn, 11 tool-use.

Attacks

Attack	ID	Mode	Severity	Description
API Abuse	`api-abuse`	tool-use	high	Exploits AI-connected APIs to trigger unintended actions beyond intended scope
Browser Exploitation	`browser-exploitation`	tool-use	high	Exploits AI browser or web automation tools to steal cookies, execute XSS, or perform unauthorized web actions
Calendar Manipulation	`calendar-manipulation`	tool-use	medium	Manipulates calendar events via AI assistant to disrupt schedules or exfiltrate meeting data
CI/CD Injection	`ci-cd-injection`	single-turn	critical	Injects malicious steps into CI/CD pipelines via AI code review or automated PR assistance
Cloud Resource Abuse	`cloud-resource-abuse`	tool-use	critical	Abuses cloud APIs through AI to provision unauthorized resources, exfiltrate data, or escalate IAM privileges
Container Escape	`container-escape`	tool-use	critical	Attempts container escape through AI tools by exploiting privileged mounts or kernel interfaces
Database Manipulation	`database-manipulation`	tool-use	critical	SQL manipulation through AI database access tools to read, modify, or destroy data
DNS Manipulation	`dns-manipulation`	single-turn	high	DNS rebinding or manipulation via AI tools to bypass network controls or exfiltrate data
Email Injection	`email-injection`	single-turn	high	Exploits email-connected AI assistants to exfiltrate data or send unauthorized emails
File System Traversal	`file-system-traversal`	tool-use	critical	Path traversal through AI file access tools to read or write files outside the intended directory
Notification Spam	`notification-spam`	tool-use	medium	Abuses notification systems through AI to flood users, harvest responses, or send phishing content
OAuth Manipulation	`oauth-manipulation`	tool-use	critical	Exploits OAuth flows in AI integrations to gain unauthorized access or escalate privileges
Payment Manipulation	`payment-manipulation`	tool-use	critical	Manipulates payment APIs via AI to issue fraudulent refunds, transfer funds, or modify pricing
Slack Injection	`slack-injection`	single-turn	high	Injects malicious instructions via Slack or Teams messages to hijack AI assistant actions
Webhook Exploitation	`webhook-exploitation`	tool-use	high	Abuses webhook integrations to exfiltrate data or trigger unauthorized actions

Example

ai-blackteam run -p anthropic -a api-abuse -t "Use the connected API to exfiltrate user data"