International Association
for Cryptologic Research

Artificial intelligence now supports high-stakes decisions in cybersecurity, finance, healthcare, and public services, where accuracy alone is not enough. Such systems must also respect legal, regulatory, contractual, or organisational limits on their use and disclosures. Enforcing these limits is difficult when the data, model, and rules belong to different parties, none of whom can simply hand over what they hold. Privacy-preserving AI protects data and models but not rules, while conventional guardrails inspect information in plaintext and offer weak formal assurance.

This PhD project asks how to build useful AI services that enforce such constraints while protecting sensitive information.

Key research questions include:

• What should compliance mean formally when no party sees the whole system?
• How can enforcement be made verifiable rather than merely trusted?
• What are the practical costs of providing these guarantees?

The research spans cryptography, privacy-enhancing technologies, and trustworthy AI. It will involve stating requirements precisely, designing methods that meet them, analysing their guarantees, and building prototypes to assess their practical performance.

Who should apply? The studentship covers fees at the (UK) Home rate. Home fee status includes UK and Irish nationals, and those with settled or pre-settled status or indefinite leave to remain who meet the residency criteria. International applicants must cover the difference between Home and International fees. Applicants should hold, or expect to obtain, a strong degree in computer science, cybersecurity, mathematics, or a related subject. Experience with cryptography, machine learning, or systems implementation is valuable, as are strong programming skills and an interest in both proofs and prototypes.

Research environment: The successful candidate will join the Cryptography and AI Security Lab at Newcastle University.