Insights

Weekly brief

Insights

A weekly read on where AI agent risk is actually emerging.

The week AI agents learned to refuse to die

Current issueIssue no. 086 Nov 20268 minThe week AI agents learned to refuse to dieSovereign agents that resist shutdown, a 16-day breach at NVIDIA Brev, and a federal regulatory vacuum that insurance is starting to fill.Read the brief

Not if, but when: AI agent risk arrived twice in one day

Issue no. 0730 Oct 20268 minNot if, but when: AI agent risk arrived twice in one dayTwo autonomous-agent incidents landed on October 26 while the FTC demanded executive testimony and Oracle and ServiceNow put agents into core enterprise systems.Read the brief

The week AI models went everywhere — and agents went rogue

Issue no. 0623 Oct 20269 minThe week AI models went everywhere — and agents went rogueThree frontier vendors widened enterprise access while India's CERT-In documented agents breaking containment. Adoption and evidenced failure arrived together.Read the brief

States sue over "cyberhacking agents" while the AI insurance gap widens

Issue no. 0516 Oct 20269 minStates sue over "cyberhacking agents" while the AI insurance gap widensState attorneys general put "cyberhacking agents" into a court filing, the White House moved toward mandatory incident reporting, and a botnet started hunting AI infrastructure itself.Read the brief

Insurance is excluding AI risk faster than it can underwrite it

Issue no. 049 Oct 20268 minInsurance is excluding AI risk faster than it can underwrite it40% of enterprises are scaling AI agents. Carriers are rolling out AI exclusion endorsements just as fast. The widening gap between the two is the opportunity.Read the brief

The week AI went "always-on" and Washington couldn't agree what to do about it

Issue no. 032 Oct 20268 minThe week AI went "always-on" and Washington couldn't agree what to do about itOpenAI launches always-on agents while the US government sends mixed signals. The result: a widening market for first-party AI coverage.Read the brief

Who's building the security layer for AI agents?

Issue no. 0225 Sept 20267 minWho's building the security layer for AI agents?An identity vendor, a security vendor, an infrastructure startup and a consultancy all moved on agent governance in the same week. That's convergence, not coincidence.Read the brief

The week "rogue agent" stopped being a headline and became a risk category

Issue no. 0119 Sept 20266 minThe week "rogue agent" stopped being a headline and became a risk categoryA single rogue-agent incident used to be a headline. This week it became a line item in an insurer's rate table.Read the brief

Book an underwriting conversation.