A grounding in three related attacks that work entirely through a machine learning model's ordinary query interface, without ever touching its weights or training data directly. Model extraction reconstructs a functionally similar copy of a model through repeated queries. Model inversion infers sensitive characteristics of the training data from a model's outputs. Membership inference determines whether a specific record was part of the training set at all. You will learn why each matters, extraction as a commercial and intellectual property concern and inversion and membership inference as direct privacy harms, and how rate limiting and query monitoring, output perturbation and confidence score limiting, and output watermarking each reduce risk without eliminating it. The course closes with a worked hypothetical investigation and a checklist for deciding, as a genuine business trade-off, how much query access and output detail an interface should expose.
Nothing. Every course, exam, and certificate on the catalog is free — including retakes. All you need is a free Safeguard account.
None. The flagship course, Safeguard Certified Practitioner, is a beginner-level course — basic familiarity with how software is built helps, but every exam question is answerable from the lessons themselves.
You can retake it after a 24-hour cooldown, as many times as you need. Retakes are free, and each attempt draws a fresh random set of questions.
The Safeguard Certified Practitioner credential is valid for 24 months from issue. The expiry date is printed on the certificate and shown live on its public verification page. Renewing means passing the current exam again.
