Carsten Maple: AI Needs Assurance Before It Earns Trust

Professor Carsten Maple of the University of Warwick and the Alan Turing Institute brought a sharp, research-driven perspective to the global conversation on AI. Maple, who has spent decades at the intersection of computer science, cybersecurity, and now AI assurance, argues that society is facing an inflection point: AI is transformative, but we still have no clear way to trust it.

He described the current moment as a wave of unprecedented hype — powered by OpenAI’s breakthroughs and amplified by accessible, consumer-friendly products that have pushed AI into the mainstream. Yet with this excitement comes an equally urgent need for confidence. “AI has the opportunity to help us enormously,” he said, “but we don’t understand how to assure it.” Regulators are struggling to keep pace, with the EU AI Act, the U.S.’s more hands-off stance, and the UK’s still-evolving approach each representing radically different philosophies.

For Maple, the foundation of trustworthy AI begins not with policy, but with science. In academia, his mission is to build the frameworks, tests, and methodologies that can verify how AI systems behave under real-world pressures. His background in cybersecurity — particularly red teaming and stress-testing systems — enables him to approach AI not as a black box, but as something to probe, break, and strengthen. “If I can break things,” he said, “we can make them more resilient.” That mindset guides his work at Verify AI, the initiative he co-founded with colleagues from the Alan Turing Institute and the University of Edinburgh to standardise assurance practices for financial institutions.

Maple’s curiosity goes back to childhood — the kind of person who dismantled VCRs to understand how they worked, sometimes reassembling them differently just to test possibilities. Today, that same curiosity is directed at AI’s emerging behaviours and unexpected impacts. He referenced examples like AI-driven executives in Japanese firms and the rapid adoption of AI-supported decision-making within financial institutions. These developments raise profound questions about governance, accountability, and human oversight.

Looking ahead, Maple believes the UK — historically a pioneer in computing and AI research — must define its role in the era of sovereign AI, where nations seek control, autonomy, and strategic capability rather than just building large language models. The UK, he argues, needs clearer government leadership and strategic investment to shape its future position.

For Maple, the challenge ahead is not simply to advance AI, but to ensure it. Only by building rigorous assurance frameworks can AI truly earn the trust required for safe deployment across finance, government, and society.