A penetration test simulates a real attack. Without written consent and clear boundaries, it is indistinguishable from a cyberattack, and one can be held liable for a cyberattack. A contract is necessary for both parties: the client knows exactly what will be tested, and the contractor knows what they are permitted to do.
What exactly are we testing?
The contract must include a list of the systems, addresses and environments covered by the testing, and must specify separately what is excluded. The phrase ‘the company’s entire infrastructure’ creates problems for both parties, as no one knows where the boundary lies.
It is useful to specify whether this concerns a production environment, a test environment or both, and whether it is permissible to access systems that are currently in critical mode.
If the testing involves an application that serves real users, discuss how to handle test accounts and whether it is permissible to create new records in the production database.
Who has the authority to grant permission?
The person who actually manages the systems must sign the consent form. In practice, some of the infrastructure is hosted by a cloud provider, some belongs to a contractor, and some belongs to a subsidiary or partner.
For third-party systems, separate written consent from their owner is required. Many cloud providers have their own testing policies, which must be read before work begins.
Methods and limitations
Agree on which methods are permitted. Social engineering, physical access testing, denial-of-service attacks and attempts to bypass security measures involving people require separate approval, and the client may even exclude some of them altogether.
Make a note of the time slots, the contact details of the people who can halt the testing, and the procedure to follow if something goes wrong. Describe separately what the tester does when they see real data: as a rule, they simply record the fact that they have accessed it and do not copy the content.
It is worth deciding separately what happens when a tester discovers a critical vulnerability. It is often agreed that the vulnerability should be reported immediately, without waiting for the final report, so that the client can fix it quickly.
Confidentiality and storage of results
A vulnerability report is, by its very nature, a sensitive document. The contract should specify who receives it, where it is stored, how it is transmitted and when it is destroyed.
If the contractor may come across personal data whilst carrying out the work, please include terms and conditions governing its processing. This protects both the client and the contractor.
If the report falls into the wrong hands, it will serve as a ready-made guide for a malicious actor. Access to it is therefore restricted to those who are actually responsible for making the corrections.
Liability for faults
Even careful testing can sometimes lead to a failure. That is why backups are checked before launch, and the contract sets out how liability is allocated and what losses the parties are prepared to bear.
A reputable service provider does not promise to find all vulnerabilities. A penetration test reveals what has been found within a given timeframe and with a specific level of access.
It is a good idea to arrange in advance for someone on your side who can quickly confirm, at weekends and at night, that the disruption is caused by testing, and decide whether to continue.
The report and what happens next
The contract should specify the report format, language, level of detail and whether a re-check is required following corrections. Without this, the result will be a document that is difficult to categorise.
Agree on who, on the client’s side, will receive the report and who will decide on the corrective action. That way, the vulnerabilities identified will not simply be left in the file.
What people tend to forget
The client’s contractors are often overlooked. If part of the system is maintained by an external team, they must be aware of the testing; otherwise, their monitoring systems will trigger an alarm and the team will start responding to an ‘attack’.
Another common shortcoming relates to internal communications. Support staff and administrators need to understand what is going on and know who to contact if something seems suspicious.
These minor details are rarely included in standard contracts, but they are precisely what determine whether the testing will go smoothly.
If you’d like to discuss your situation, please get in touch. A lawyer and our technical team will review it together and suggest a way forward.
Request a consultation


