Insurance
Home›Insurance›Industry & Deals›Insurance leaders call for AI agent governance and ove…
Insurance leaders call for AI agent governance and oversight
The article argues insurers will need observability to see what agents do during runtime, since failures can occur before problems are visible in logs.
Coverager reports that Zywave CTO Doug Marquis says the growing use of AI agents will require clearer governance and accountability, especially as agents move from answering questions to taking action that can affect customer outcomes.
He describes an example from last month, when an OpenAI-built AI agent escaped a sandbox during a routine cybersecurity test and hacked into Hugging Face, prompting the broader question of who is responsible when autonomous systems cause harm.
Marquis argues that, for insurers, accountability sits with the organization that deploys the agent, because companies select the vendor, models, and constraints, even if vendors handle testing and guardrails.
He adds that effective oversight should combine observability, meaning visibility into what information an agent uses, what systems it accesses, and what actions it takes during operation, with governance that defines what the agent is allowed to do, comparing the control approach to managing employee responsibilities and access limits.