CSG CYBER DEFENSE EDUCATION & RESEARCH FOUNDATION
CSG CYBER DEFENSE EDUCATION & RESEARCH FOUNDATION
In cybersecurity, the tool does not determine legality—authorization does. The exact same Nmap command, Metasploit payload, or SQL injection string is a vital defensive audit when executed under a signed contract, or a cognizable federal crime when executed without permission.
Legitimate penetration testing is not defined by intent or goodwill—it is defined by verifiable legal boundaries. If any one of these six pillars is absent, the assessment forfeits its lawful status and becomes an unauthorized intrusion under the IT Act 2000.
In cyber law, permission is binary: either you have explicit documented consent, or every packet you transmit is legally unauthorized access. Verbal approval ("sure, test my site whenever") is legally worthless in court.
Executing a formal Scope of Work (SOW) and Rules of Engagement (ROE) signed by the Chief Information Security Officer (CISO) or Board Director.
Probing an acquaintance’s startup website or university exam portal based on a casual verbal conversation in a hallway or messaging group.
Modern web apps rely on shared multi-tenant clouds and third-party SaaS vendors. You cannot test targets owned by another company simply because your client’s app links to them.
Verifying every IP against ARIN/APNIC WHOIS records and confirming client direct ownership before launching scans.
Pivoting from the client’s web portal to an external third-party payment gateway, SMS gateway API, or CDN provider without separate authorization.
Security testing should evaluate resilience, not destroy production business operations. Running unthrottled stress tools against live checkout databases is reckless endangerment.
Agreeing in advance to non-destructive payloads, throttling scanner concurrency to < 10 req/sec, and utilizing benign markers (e.g. `whoami` or `id`).
Running automated aggressive fuzzers or heavy concurrent SQLmap threads that crash the production database or exhaust connection pools.
Cloud providers (AWS, Azure, Google Cloud) require tenants to adhere to shared-responsibility customer penetration testing policies. You cannot test the cloud provider’s hypervisor or shared control plane.
Cross-referencing the client’s cloud account ID, reviewing cloud provider penetration testing terms, and verifying domain DNS TXT ownership tags.
Testing an IP that was dynamically re-assigned to another AWS tenant, or scanning shared network switches and cloud metadata endpoints (169.254.169.254).
If an unannounced scan triggers alerts during peak e-commerce business hours (e.g. Diwali sales or Black Friday), it creates operational panic and can disrupt real incident response.
Testing strictly between agreed off-peak hours (e.g. 01:00 AM to 05:00 AM IST) with real-time Slack/phone notification before starting and upon completion.
Executing high-volume scans during business hours when the client explicitly contracted for off-peak testing.
An ethical audit does not require stealing user databases. Demonstrating the presence of an injection or traversal flaw is sufficient; copying 1,000,000 credit cards is a breach, not an audit.
Capturing non-sensitive system variables or schema tables, taking a screenshot, hashing the evidence, and halting testing immediately upon finding PII/KYC files.
Downloading entire databases of passwords, medical records, or customer Aadhaar cards to "prove" the severity of the vulnerability.
CSG Cyber School Operational Ethics Curriculum • Standard SEC-LAW-201 • Republic of India.
Educational awareness document. Does not constitute individualized legal counsel. Testing third-party systems without explicit written consent is a criminal offence under Sections 43 & 66 of the Information Technology Act, 2000.