21.07.2026

Zero Trust for AI: Securing Intelligent Systems in a New Era

WEBINAR: 2pm Thursday 23rd July 2026

About our webinar presenter Prashant Vajpayee: Prashant is Senior product leader, AI researcher, and Global Fellow (AI2030) with 20 years of experience in enterprise data governance, secure systems integration, agentic AI, and cybersecurity. Recognized for bridging advanced research with enterprise‑scale implementation, delivering trusted, compliant, and AI‑ready data ecosystems.

About our webinar: Zero Trust for the AI Era. Prashant helps us explore and understand how the principle of “never trust, always verify” must be extended to AI systems, ensuring every model action, data request, and tool interaction is continuously authenticated and risk‑evaluated. As AI becomes increasingly autonomous, Zero Trust provides the essential safety layer that prevents hallucinations, misuse, and unauthorized access from causing real‑world harm. This approach transforms AI workflows from static approvals into continuous, real‑time trust assessment, enabling safer and more resilient AI adoption across modern enterprises.

How to join: Zero Trust for the AI Era Webinar. Members of the IAP member will have received an email with a Teams meeting link. If you would like to join the webinar but are unable to find the link or if you have a question about the webinar email admin@iap.org.uk.

What Is Zero Trust for AI?

As artificial intelligence becomes embedded in business operations, the traditional approach of trusting users and systems inside a network is no longer sufficient. Zero Trust for AI applies the core principle of “never trust, always verify” to AI applications, models, data, and users. Every request is authenticated, authorised, and continuously validated, regardless of where it originates. This reduces the risk of data breaches, model manipulation, and unauthorised access.

What is the Approach?

AI systems often process sensitive information and interact with multiple data sources, making them attractive targets for cybercriminals. Threats such as prompt injection, data poisoning, model theft, and credential compromise highlight the need for stronger security controls. A Zero Trust approach helps organisations protect AI assets by enforcing least-privilege access, monitoring user behaviour, encrypting sensitive data, and validating every interaction with AI models.

The Zero Trust AI Strategy

An effective strategy starts with identifying and classifying AI assets, including models, datasets, APIs, and infrastructure. Organisations should implement strong identity management, multifactor authentication, continuous monitoring, and automated threat detection. Regular model testing, governance policies, and secure development practices further strengthen resilience. Combining Zero Trust with AI governance ensures security and compliance evolve alongside rapidly changing AI technologies.

Key Developments in the UK

Zero Trust for AI is becoming a key part of the UK’s cyber security strategy as government, industry and security leaders respond to the rapid rise in AI-enabled cyber attacks. The approach is based on the principle of “never trust, always verify”, ensuring that every AI model, user, application and data exchange is continuously authenticated and granted only the minimum level of access required. This helps reduce the risk of unauthorised access, data theft and the misuse of AI systems.

The UK government is incorporating advanced AI capabilities into its National Cyber Shield programme to strengthen cyber defence. The initiative promotes transparent, auditable AI systems and a federated Zero Trust architecture designed to protect critical national infrastructure.National Cyber Security Centre (NCSC) guidance has highlighted that AI is significantly reducing the time between the discovery of software vulnerabilities and their exploitation by cyber criminals. As a result, organisations are encouraged to adopt a Zero Trust mindset, assume that breaches can occur, and implement robust identity, authentication and least-privilege access controls. By extending Zero Trust to AI technology providers are evolving traditional Zero Trust security frameworks to protect the entire AI lifecycle. This includes safeguarding training data, securing data ingestion pipelines, validating models before deployment and monitoring the behaviour of AI agents throughout their operation. By strengthening Data Governance, as AI-generated content becomes increasingly widespread, organisations are placing greater emphasis on trusted data sources. Verifying the origin and integrity of training data helps prevent model degradation, improves the reliability of AI systems and supports compliance with UK data protection and governance requirements.

New tools and guidance from Microsoft: microsoft.com/security/announcing-zero-trust-for-ai/