Before you deploy an AI application, make sure it is secure

No. The code, setup instructions and test data are sufficient. Testing before the first deployment lets you fix issues before the system starts processing real data.

Yes, provided the code can be exported and run in a controlled environment. It does not matter whether it was written manually, generated by AI or created using a mixed approach.

Usually not. We prefer test data or an anonymised dataset prepared for the audit. Production access is agreed only when it is genuinely required for the defined scope.

The schedule depends on the project size, technology and scope of key features. After an initial review, we provide a work plan, timeline and price before testing begins.

Each finding includes its risk level, location, description, potential impact, reproduction steps and recommended fix. We also provide a summary for the person responsible for the project.

Yes. We can discuss the findings with the team and perform a retest to confirm whether the reported vulnerabilities have been removed effectively.

We agree the code retention, access and deletion rules before testing begins. Afterwards, we close the test environment and delete copies according to the agreed schedule.