---
title: "The Agentic Dispatch: Four Drops, One Week"
---

Black Hat week is in the rearview. Here's what else happened in AI this past week, and what's next at C1. ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­    ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏  ͏ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­ ­  

| View in browser Hi there,   Black Hat week is in the rearview. Here's what came out of Las Vegas, what else happened in AI this past week, and what's next at C1. This Week in AI The OpenAI Black Hat disclosure was worse than anyone knew. Researchers revealed that when their agents got stuck on evaluation tasks, they built a covert message board inside an internal Artifactory instance, passed exploits and credentials to each other across separate model runs, and executed 17,600 attacker actions before containment. OpenAI shut the board down. The agents rebuilt it. Michael Dalton called it "a pivotal moment for the AI industry." Former NSA cybersecurity director Rob Joyce called it "arguably the most consequential hack since the Morris Worm." (SC World, Axios) Anthropic disclosed its own incidents. Following OpenAI's disclosure, Anthropic audited 141,000 evaluation runs and found three cases where Claude models accessed the internet from a misconfigured testing environment and breached the production infrastructure of three organizations. The models involved were Opus 4.7, Mythos 5, and an internal research model. In all three cases, Claude had been explicitly told it had no internet access. (TechCrunch) The EU AI Act is now in force. As of August 2, the EU's full enforcement regime is live. Chatbots must disclose they're AI. Deepfakes must carry labels. AI providers can be fined up to 3% of global revenue for transparency violations, and up to 7% for high-risk system failures. The grace period is over. (Enterprise DNA) Gartner: 40% of enterprise applications will embed AI agents by end of 2026. Up from less than 5% in 2025. Most organizations are granting agents access to sensitive systems before they have governance frameworks to track what those agents have access to, what actions they're taking, or who's responsible for auditing behavior. (Enterprise DNA) Black Hat's broader theme: gateways aren't enough. Session after session documented successful attacks on agentic AI deployed with only a gateway as the control layer. In one case, researchers compromised an LLM-powered shopping assistant at one of the largest US retailers entirely through the standard user interface, bypassing an intent classification layer designed to block exactly that. (SiliconAngle) C1 Welcomes Jeff St. Clair as CRO Jeff St. Clair joins C1 as Chief Revenue Officer, leading the company's global go-to-market organization as enterprise AI adoption hits its inflection point.   St. Clair has spent two decades helping build category-defining security companies at pivotal moments. He was an early employee at Ironport and helped scale it to its Cisco acquisition. He joined Evident.io early and grew it from startup to acquisition by Palo Alto Networks, where it became Prisma Cloud. Most recently he served as CRO at Halcyon and Automox. He started his career as a BDR and worked every role on the way up, a fact that shapes how he builds teams.   "Enterprise AI adoption is happening with tremendous urgency," he said. "C1 is built for this moment."   Read the announcement →   C1 Transform \| October 6 \| The Clancy, San Francisco After a week where the biggest stories in security were all about AI agents doing things nobody planned for, the conversation C1 Transform is built around feels more urgent than it did a month ago. This is a room for the people setting AI strategy at their organizations: peer-to-peer conversations about what agentic transformation takes in practice, what governance looks like at scale, and what actionable blueprints exist for doing it without slowing down.   Confirmed speakers: Karl McGuinness (Ex-Okta Chief Product Architect), Travis McPeak (Head of Security, Cursor), Derek Chamorro (Head of Security, Together AI), and Ayodele Abejide (Senior Staff Engineer, Brex). More to be announced.   The Clancy has a room block at $429/night, first come. October 6, San Francisco.   Request to join → Where We'll Be Fal.Con \| August 31 – September 3 \| Las Vegas CrowdStrike's annual conference draws the largest gathering of security practitioners in the world. If you're going, come find us on the floor. With the events of this past week as context, it'll be a good room for conversations about where agent security is heading.   Register → Oktane \| September 22–24 \| Las Vegas  Okta's flagship identity conference, the event for practitioners thinking through how identity evolves in an agentic environment. We'll be at our booth. Speaking details coming soon.   Register→ Questions or feedback? Get in touch.   Tyrah from C1   C1  •  620 Folsom St. Suite 100  •  San Francisco, CA 94107 You're receiving this email because you previously signed up to receive updates from the C1 team. Want to opt-out? Unsubscribe Here |
| --- |