Beyond existential fears, AI presents an immediate challenge: it dramatically lowers the barrier for enacting harm, acting as a powerful "co-creator of evil". This session confronts this uncomfortable reality, moving beyond simplistic guardrails to explore AI's role in amplifying both intentional malice and systemic failures. What happens when the same AI boosting workplace productivity and executive function assists those with harmful intent?

Drawing on frontline experience, including red teaming AI systems for Chemical, Biological, Radiological, and Nuclear (CBRN) threat scenarios, Angus Allan (AI Product Leader & Scottish AI Alliance Governance Group) examines how AI can assist in carrying out complex, harmful actions—tasks that previously required rare expertise. We'll explore how generative and agentic systems are changing the dynamics of misuse, and how flawed systems deployed at scale can unintentionally magnify societal harms in corporate and public sectors.

Key Takeaways:
- Understand the "Co-creator" Risk: Recognise how AI's assistive nature lowers the capability threshold for executing harmful acts.

- Learn from High-Stakes Testing: Gain unique insights derived from assessing AI capabilities in critical threat scenarios (like CBRN) and understand the limitations of current safeguards.

- Implement Responsibility by Design: Acquire actionable strategies and design principles to proactively build AI systems that resist co-creating harmful outcomes.