Exploring AI Network Protocols; Vulnerability Truths and Guarantees; and the News – O’Shea Bowens, Jeremiah Grossman – ESW #469
Segment 1 - Interview with O'Shea Bowens
What do we really know about "AI Network Protocols"? Network security is about to get popular all over again.Generative AI caused a disruptive explosion across all of tech and every company’s roadmap. The move from chatbots to AI agents doubled down on that disruption. Now agents need to talk to each other?Boom: we have MCP. A2A. Universal Commerce Protocol. General purpose and specialized protocols for agent communication. What does this look like from the network perspective, though? O’Shea Bowen joins us to answer this question, and he thinks the results are interesting enough to spark a resurgence of interest in network security tooling.Segment Resources:- https://www.nsa.gov/Portals/75/documents/Cybersecurity/CSIMCPSECURITY.pdf?ver=bmgiSbNQLP6Z_GiWtRt6bg%3D%3D
- https://labs.cloudsecurityalliance.org/research/csa-research-note-mcp-security-crisis-20260504-csa-styled/
- https://cyberone.security/blog/building-an-ai-security-strategy-without-stalling-business-growth
Segment 2 - Interview with Jeremiah Grossman
Jeremiah Grossman on why we've been measuring cyber risk wrong for 20 yearsAfter decades helping shape modern web security, and building companies that were ultimately acquired by Synopsys and Tenable, Jeremiah Grossman believes cybersecurity has arrived at an inflection point. His argument is a provocative one: for years, the industry has optimized around the wrong metrics. His latest venture, Root Evidence, aims to help security teams identify which risks are most likely to cause meaningful business loss, and he has the evidence - real-world breach data, cyber insurance claims, digital forensics intelligence, attack surface intelligence, and observed attacker behavior - to back it up.Find all of CyberRisk TV's Black Hat 2026 coverage at: https://www.securityweekly.com/blackhatSegment 3 - Weekly Enterprise News
Finally, in the enterprise security news,- We vibe check the AI model situation
- hidden devices in California cars causes concerns
- OpenAI’s models escape sandboxes and breaches another AI company, totally by accident, they promise!
- Grok Build uploads all your files, totally by accident, they promise!
- Eclipsium debuts a firmware version of patch tuesday!
- HTTP gets a new method
- common problems with incident response
- Which one of the security weekly hosts would consider switching to a “dumb phone”?
O’Shea Bowens is a 16-year cyber security practitioner. He’s a hands-on leader in the areas of security program management, incident response and threat hunting. He’s currently researching the intersection of artificial intelligence and cyber security with a focus at the network layer. He considers himself a jack of trades rather than a master of one and this approach has created many adventures during his career. In other areas of life: He’s father to an amazing 8 year old daughter, supporter and player for North Shore Rugby Club, and an avid snowboarder
Jeremiah Grossman is CEO of Root Evidence and a pioneering cybersecurity leader with over 25 years of industry impact. He was one of Yahoo’s first security officers and went on to found WhiteHat Security and Bit Discovery, acquired in just three years. He now invests in cybersecurity startups through Grossman Ventures and continues to shape modern web application security and attack surface management.
- Security leaders, your vulnerability program is overloaded. Thousands of findings, limited resources, and no clear way to prioritize what actually matters to the business.Meanwhile, regulators and boards expect measurable risk reduction, not just scan results.Join the Vulnerability Management Virtual Cybersecurity Summit on July 29th to learn how leading organizations are shifting from volume to risk-based prioritization and turning exposure into actionable strategy.Security Weekly listeners can register for free at https://securityweekly.com/vulnmanagement using the promo code: CSS26-SW
- InfoSec World brings cybersecurity professionals together across industries, from healthcare and financial services to government and the Fortune 500. Join the community in Orlando, October 12–14, for practical education, new perspectives, and cybersecurity research unveiled live. Listeners save 30% on their pass with code ISW26-SWSAVINGS at securityweekly.com/infosecworld2026.
Adrian Sanabria
- FUNDING/M&A, courtesy of the Security, Funded newsletter, issue #251 – Silence of the M&As
VIBE CHECK
What does the future of AI & Cyber look like?
- 60% - Local open-source models
- 20% - Closed-source US models
- 20% - Open-source Chinese models
- 0% - Something else (tell me)
But... aren't Chinese open-source models just a subset of local open-source models?
FUNDING
- Oak, an Israel-based AI-native identity operating system for enterprise access management, raised a $60.0M Seed from Accel, CRV, and Greylock.
- Valarian Technologies, a United Kingdom-based secure communications infrastructure platform, raised a $50.0M Series A from New Enterprise Associates.
- Risk Ledger, a United Kingdom-based third-party software supply chain security platform, raised a $32.2M Series B from Axiom Equity.
- Beacon Security, an Israel-based security data pipeline and orchestration platform, raised a $13.0M Seed from Notable Capital.
- Casco, a United States-based AI-driven red-teaming for web apps, APIs, and AI systems, raised a $13.0M Series A from Standard Capital. <- not many US-based companies getting funding this week!
- Pulse Security AI, a United States-based AI-driven cybersecurity operational program management platform for security leaders, raised an $8.0M Seed from Foundation Capital. <- why would you name it Pulse Security???
ACQUISITIONS
- CardinalOps, an Israel-based AI-powered security engineering platform, was acquired by Cribl for an undisclosed amount. CardinalOps had previously raised $24.0M in funding.
- VULNERABILITIES: A Device Hidden in Cars Across the US Leaves Them Vulnerable to Hacking and Paralysis. Patch It Now
This is a nightmare situation for car owners in SoCal. What a mess.
- BREACHES: OpenAI Says Its AI Models Escaped Sandbox, Targeted Hugging Face to Cheat Benchmark
OpenAI, arguably the company with the longest and deepest experience sandboxing and testing LLMs just couldn’t keep this one contained ¯_(ツ)_/¯
What a great coincidence it all resulted in a partnership and all worked out for the best!
OpenAI's announcement
- BREACHES: Grok Build Uploaded Entire Git Repositories to xAI Storage, Not Just Files It Read
Does this count as a breach, when your AI programming tool steals all your files? I think it does.
All the breaches we're talking about this week were caused by AI tech companies. Weird. Coincidence?
- RESOURCES: New InfraTrust report reveals infrastructure flaws admins should patch first
Eclipsium's new InfraTrust report is basically Patch Tuesday for firmware updates!
About time we had a resource for such an overlooked area.
- NEW FEATURES: HTTP is getting a new method called QUERY (Instagram: @umacodes)
"HTTP QUERY is designed to fill a gap that GET and POST never handled cleanly: searching for data with complex filters, sorting, pagination, arrays, and nested conditions."
Security folks might not be familiar with much more than GET, POST, and TRACE, of course, since it shows up on every SSC/Bitsight naughty list, dropping our grades to a C or D :sweat-smile-emoji:
We don't often get new HTTP methods, so it's worth getting familiar with it!
- REGULATIONS: The CMMC Pause: The DoD Suspends Phase 2 Requirements – Embedded
The latest on the CMMC phase 2 pause.
Some great in-depth commentary and recommendations from James Goepel as well.
- RESEARCH: Common Problems Plague Our Incident Response Operations
A great modern analysis on the state of IR and its problems/challenges.
- SQUIRREL: Commodore Made a Digital Detox Phone That Isn’t Dumb
- SQUIRREL: Light Flip
Ayman Elsawah
- Balancing AI Security In The Enterprise
AI Security in the enterprise is evolving every single day. In this event we will be speaking to an expert that has been at the forefront of securing AI in the enterprise. We will talk about some of the biggest challenges facing the enterprise today.
Topics include: - How do we enable employees to use AI safely? Exploring the balance between productivity and security. - What is Zero Data Retention (ZDR)? Limitations with Claude Fable and Grok - Shadow and Rogue AI Agents - Can we really protect our data from leaving the building with AI?
