Lionel Tarassenko
Crossbench life peer and Oxford engineering professor specialising in machine learning for healthcare.
Lord Tarassenko wants the AI Security Institute to audit frontier AI before it runs inside regulated systems. His amendment to the Cyber Security and Resilience Bill faces The House of Lords report stage on 26 October 2026.
Last refreshed: 12 October 2026 · Appears in 1 active topic
Should frontier AI be audited before it runs inside regulated national systems?
Timeline for Lionel Tarassenko
Tabled AI Security Institute audits of frontier AI
Cybersecurity: Threats and Defences: Peers table AI amendments to cyber billBackground
Lionel Tarassenko has sat in The House of Lords as a crossbench life peer since June 2024. He joined the Lords Communications and Digital Committee in January 2026.
At the University of Oxford he holds a chair in engineering Science. He headed the Department of Engineering Science from 2014 to 2019 and is the founding President of Reuben College. His research applies signal processing and machine learning to healthcare.
Lord Tarassenko tabled an AI audit amendment to the Cyber Security and Resilience Bill ahead of its 26 October 2026 report stage. The bill extends Britain's cyber obligations past today's regulated operators to the IT firms that serve them, and peers added managed service providers and data centres at second reading. His report-stage amendment would require the AI Security Institute to audit frontier AI products before they are used in regulated systems.
His audit amendment reaches report stage
Lord Tarassenko has tabled an amendment that would require the AI Security Institute to audit frontier AI products before they are used in regulated systems. It sits among the AI amendments peers tabled before the bill's Lords report stage on 26 October 2026.
The proposal draws directly on his own field. An Oxford engineer WHO applies machine learning to healthcare, he is asking for a pre-deployment check on the kind of system he has spent his career building. Other amendments in the same batch target AI developers' reporting of hacks and ministerial shutdown powers; his is the one that puts the audit before the system goes live.
