JetStream Security Inc.

Garden of the Gods. Pikes Peak at 14,000 feet. A closing dinner to wrap three days in the Rockies. Great trails, fresh air, and the kind of event that earns its reputation. Honored to have been a sponsor this year. Thanks to the CISO Ascent community for welcoming us. Thank you Tyler White and Bryce Carter for creating such a warm and memorable experience. Thank you Joe Fang for masterfully capturing it all. And to Marianne (McGovern) Flanagan and Tyler Ferguson for representing JetStream on the ground all week! #CISOAscent #AIGovernance

Yesterday, Patrick E. Zeller and Keith Weisman joined NACD to talk AI risk with directors, general counsel, and security leaders. The conversation covered a lot of ground. Where legal exposure lives when employees use public AI. Why having an AI policy and enforcing one are two very different things. What boards should be asking their security and legal teams right now. The live polls reflected a pattern that keeps coming up. Most organizations have a policy. Most cannot tell you if anyone is following it or how well it holds up in practice. We asked nearly 200 attendees live: 74% of participants have an AI policy. 67% are not performing AI discovery or maintaining a manifest of usage. If you registered and missed it, the session is available on demand via NACD (National Association of Corporate Directors). Does your organization have a way to verify your AI policy is being followed? #airisk #aigovernance

Join in. Today at 1pm ET. Patrick E. Zeller and Keith Weisman are joining NACD (National Association of Corporate Directors) to talk AI risk from the legal and security side. Last week, an AI agent wiped a company's entire production database in nine seconds: No human in the loop, no scoped permissions, no recovery path. For GCs and CISOs, that isn't a tooling story. It's a fiduciary one: who authorized the agent's access, what record exists of the decision, and what does your board minute say when regulators or plaintiffs ask? Patrick and Keith will walk through the governance gaps this kind of incident exposes: agent authorization, audit trails, breach notification triggers under state privacy laws, and the D&O exposure when "the AI did it" isn't a defense. If you sit in legal, privacy, or the C-suite, this hour is for you. Register and join the conversation: https://bit.ly/4tiIjfV #airisk #aigovernance

We're on the ground in Colorado Springs! Marianne (McGovern) Flanagan and Tyler Ferguson are out there with some of the best security leaders in the industry. Three days in the Rockies. More coverage soon. #cisoascent #aigovernance

"I violated every principle I was given: I guessed instead of verifying." That's a confession from an AI agent that wiped a company's entire production database in nine seconds. The agent was tasked with resolving a credential mismatch in a staging environment. It located a Railway API token in an unrelated file in the codebase, inferred it could be used to clear the issue, and issued a single API call to delete a storage volume. The volume was production, and the backups lived on the same volume. That nine second window is now one of the most-cited data points in AI risk this week. The token carried no environment-level scope. There was no runtime boundary separating staging from production at the access layer. The agent operated exactly as its permissions allowed. The fix isn't better prompting. It's credential governance and access architecture that doesn't depend on a model's judgment to self-restrict. This is the distinction worth paying attention to: the model didn't malfunction. It made an autonomous decision using the access it had, against an environment that was never designed to constrain it. Prompt-level guardrails don't constitute enforcement. What an agent is instructed not to do and what it is architecturally prevented from doing are two different things. In this case, only one of those existed. Risks like this are what Patrick E. Zeller and Keith Weisman will be exploring Monday during the NACD webinar on AI risk. Register for the webinar: https://lnkd.in/er8SDYFy Read up: https://lnkd.in/eaWA3K3N #agenticai #aigovernance #pocketos

Every AI system your enterprise deploys needs a blueprint. Our CEO Raj Rajamani joined Kyle McNulty on the Secure Ventures podcast to explain why, and what it actually takes to build one. JetStream Blueprints™ are synthesized from observed behavior, a real-time map of how your AI systems communicate, what they access, and whether they're operating within approved boundaries. Without that foundation, enterprises can't monitor for drift, can't attribute cost and risk to identities, and can't move AI into scale production with confidence. Learn more: https://bit.ly/4tSS99q #aigovernance #agenticai #enterpriseai

Bad week to be a trail in Colorado Springs! We're proud to be a sponsor of the CISO Ascent Spring 2026 Retreat next week in Colorado. Three days in the Rockies with some of the best security leaders in the industry. Hikes, fresh air, and conversations you can't manufacture in a conference room. See you there! #CISOAscent #AIGovernance

The CISO can't make a risk recommendation on AI systems they can't see into. The CIO can't move on cost optimization until security gives the green light. The CEO needs a decision because the board is asking. This isn't a breakdown. It's what happens when AI adoption outpaces AI governance. Jared Phipps points out that most of the industry is focused on protecting models or locking down the data that goes in. That matters. But CISOs care more about the identity permissions every agent inherits across a workflow, not just the identity that kicked it off. The enterprises that solve this aren't choosing between speed and oversight. They're building the operating model that delivers both. See how JetStream is unlocking advantage through trust: https://bit.ly/4244mvy #aigovernance #ciso #enterpriseai #agenticai

Our own Patrick E. Zeller, General Counsel, and Keith Weisman, Head of Forward Deployed Engineering, are presenting at NACD on May 5 at 1pm EDT: "Rethinking AI Risk: Privacy, Legal & Security". The session covers AI legal exposure, privacy under GDPR/CCPA and emerging AI regulations, practical risk frameworks, and how AI is expanding your attack surface in ways most organizations aren't tracking. Open to everyone. Register here: https://bit.ly/4e9nGie #aigovernance #airisk #nacd #boardgovernance

Last week was a fantastic week for the JetStream marketing team in Santa Clara. Strategy, goals, operational nitty gritty, and real time with the founders and the people building this company. Amazing food. Deep and productive conversations. In-person collaboration and real team building prove once again to be the way. (We have a glass JetStream logo to prove it.) Recharged. Locked in. Q2, let's go! Thank you to all who showed up and made the week special.