Computing Library › Security & Safety-Critical Computing
Security & Safety-Critical Computing

Penetration Testing and Red Teaming

Authorized simulated attacks that test defenses in practice, revealing what real adversaries could actually accomplish.

Testing Defenses by Attacking Them

Reasoning about security is not the same as proving it. Penetration testing authorizes skilled testers to attack a system as an adversary would, within agreed rules, to find exploitable weaknesses before real attackers do. It answers a question that documentation cannot: not are we supposed to be secure, but can someone actually get in, and how far.

Penetration Test vs Red Team

Kronos motion — 14 mev materials test

What a Red Team Tests

A red-team exercise emulates a realistic adversary pursuing a specific objective, reaching a critical system, exfiltrating defined data, and measures not only whether they succeed but whether the defenders notice and respond. It stresses people and process, not just technology, revealing gaps in monitoring and incident response that a narrow scan would never surface.

Scope and Safety

On operational and safety-critical systems, live attack testing carries its own risk: a test that disrupts a running process is unacceptable. Such systems are typically tested against representative replicas, digital twins, or test benches, or with carefully bounded, non-disruptive techniques, so the assessment never endangers the real plant.

Fusion Context

Security testing for a fusion program favors representative environments over the live machine, so assessment never risks the process or the safety case. For the Hyperion breeder and burner designs, threat models are validated against test benches and simulation environments during the current design phase, ahead of the Q2 2027 construction start, where weaknesses can be found and fixed without any consequence to real hardware.