What are the main security risks of using AI in government?
Answered
AI-specific risks include data/model poisoning, data leakage, insecure AI tooling, prompt injection, perturbation attacks and hallucinations, as well as amplification of existing cyber risks.
From the guidance
Primary (how) AI Playbook for the UK Government
Some threats – such as data poisoning, perturbation attacks, prompt injections and hallucinations – are specific to AI. However, AI systems can also amplify generic risks such as phishing and cyber attacks.
Section: Principle 3: You know how to use AI securely
Read this in AI Playbook for the UK Government (opens in new tab)
Related questions
- Do staff need AI-specific cyber security training? Delivery, assurance and operations
- Who is responsible for security when we use third-party AI components? Buying and building
- What is data poisoning in an AI system?
- What is adversarial machine learning and why does it matter?
- What is the difference between prompt injection and jailbreaking?