Skip to content
AI Security

UK AI Security Institute Adversarial ML & Model Privacy interview questions

Adversarial ML & Model Privacy is a core part of the UK AI Security Institute AI Security Engineer loop. The attacker model for ML, evasion and adversarial examples, availability, targeted and backdoor poisoning, poisoning at web scale, model extraction, membership inference, training data extraction, model inversion, differential privacy and privacy auditing. Questions tagged to UK AI Security Institute come first. After them, the top-voted questions from the Adversarial ML & Model Privacy track. Each answer reasons the way a senior security engineer would in front of a panel: attacker, asset, control, and where the control breaks.

Adversarial ML & Model Privacy questions tagged to UK AI Security Institute

3 questions · 0 unlocked for you
Open full answer →
Open full answer →
Open full answer →

More Adversarial ML & Model Privacy questions for UK AI Security Institute's loop

Sorted by reader votes. These are not tagged to UK AI Security Institute; they cover the adversarial ml & model privacy ground its AI Security Engineer loop is known to test.

15 questions · 5 unlocked for you
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →
Open full answer →

Concepts behind UK AI Security Institute's Adversarial ML & Model Privacy round

The background these answers assume. Introductory lessons are free; advanced ones need Premium.

Foundational
Adversarial Examples and EvasionAn adversarial example is an input deliberately perturbed, often imperceptibly, so that a model makes a wrong prediction. Evasion attacks use them at prediction time against a fixed model, with or without access to its internals, and they often transfer between models. Defences such as adversarial training and certified robustness help within limits, and many published defences have been broken by adaptive attacks. The same idea appears in language models as optimised adversarial suffixes.
Foundational
Data PoisoningData poisoning manipulates training, fine-tuning or retrieval data so a model behaves as the attacker wants: degrading accuracy, causing targeted mistakes, or planting a hidden trigger. Models trained on scraped web data, user feedback or third-party datasets are exposed because attackers can influence those sources. Research has shown that poisoning web-scale datasets is practical and that small numbers of documents can implant backdoors in large language models. Defences rest on data provenance, curation, anomaly detection and evaluation, and none is complete.
Foundational
Membership Inference AttacksA membership inference attack tries to determine whether a specific record was part of a model's training data, using the model's outputs. Models behave measurably differently on data they were trained on, especially when they overfit, and an attacker can exploit that gap. For sensitive datasets, membership itself is private information. Differential privacy is the principled defence; regularisation, limiting outputs and data minimisation reduce risk.
Foundational
Training-Data Memorisation and ExtractionLanguage models memorise part of their training data verbatim, and an attacker can sometimes make them emit it: personal data, secrets, copyrighted text. Memorisation grows with duplication, model size and prompt context, happens before overfitting, and survives alignment, which only hides it. Defences stack: deduplicate, scrub, train with differential privacy where it fits, filter outputs knowing approximate copies slip through, and measure with canaries and extraction tests. Fine-tuning on customer data is the highest-risk case.
Foundational
Model ExtractionModel extraction uses query access to a model to steal its functionality or parameters: training a substitute that imitates it, or recovering specific internals. It lets attackers avoid training costs, bypass usage terms, and build better attacks against the original. Rich outputs such as probabilities and log-probabilities make it easier. Defences combine rate limits, restricting output detail, monitoring for extraction-like query patterns, contractual terms and, where useful, watermarking.
Core
Sign in
Backdoors and Trojaned ModelsA backdoored model behaves normally except when a specific trigger appears in its input, at which point it does what the attacker chose. Backdoors are implanted through poisoned training or fine-tuning data or by directly modifying weights, and they reach victims through shared pre-trained models and datasets. They are hard to find because testing rarely includes the trigger, and research has shown some backdoors survive later safety training. Provenance, controlled fine-tuning, evaluation and architectural containment matter more than detection.
Explore the full curriculum →
UK AI SECURITY INSTITUTE ADVERSARIAL ML & MODEL PRIVACY FAQ
What Adversarial ML & Model Privacy questions does UK AI Security Institute ask in interviews?▲

UK AI Security Institute's AI Security Engineer loop draws adversarial ml & model privacy questions such as "What is gradient masking, and why is it a trap when you evaluate defences?", "A team claims their model is now robust to adversarial inputs. How do you evaluate that claim honestly?", "How much poisoned data does an attacker need, and why does that matter at web scale?". The attacker model for ML, evasion and adversarial examples, availability, targeted and backdoor poisoning, poisoning at web scale, model extraction, membership inference, training data extraction, model inversion, differential privacy and privacy auditing. Every question in the topic is listed below from easiest to hardest, each with a worked answer.

How should I prepare for the UK AI Security Institute Adversarial ML & Model Privacy round?▼
Does UK AI Security Institute hire AI security engineers?▼
What is the UK AI Security Institute AI security interview process?▼

Other UK AI Security Institute interview rounds

The other tracks UK AI Security Institute's AI Security Engineer loop tests.

UK AI Security Institute ai red teaming→

Prep the whole UK AI Security Institute AI Security Engineer loop

Adversarial ML & Model Privacy is one round. Unlock every answer across UK AI Security Institute's full loop with Premium. Free questions in every track let you try the depth first.

Every worked answer in all 17 topics, every lesson in every course, the full concept curriculum, the complete practice-test bank, and the downloadable security field guides.

6 months · One payment · No auto-renewal

Unlock all answers · $25Full UK AI Security Institute interview guide

Independent and not affiliated with UK AI Security Institute. All trademarks belong to their owners.