In this interview with The Insurer, Anthony Cordonnier, Global Co-head of Cyber, and Antonis Papsavva, Cyber Threat Intelligence Expert, discuss how participants in the cyber market still need to be wary going forward, despite a lack of claims tied to emerging attacks on AI systems.
Anthony said that losses from new adversarial tactics targeting AI models and agentic systems have yet to trigger claims, but he expects that to change.
“A lot of this is yet to translate into claims, and I think it wouldn’t give us a false sense of comfort,” Anthony said. “These will be coming.”
Anthony’s comments follow a recent report that found the attack surface for AI has expanded, particularly for agentic systems that can ingest external content, retain memory and invoke tools.
Antonis indicated that these attacks are not necessarily a whole new family of cyber threats, but rather an adaptation of existing attack methods to large language models and AI agents.
“It is something that the industry needs to keep an eye on, because potentially they could be scaled and have a wider effect,” he added.