EP 36 — ruby's George Al-Koura on why 15 certifications still won't save you in a live SOC scenario
Future of Data SecurityShow Notes
George Al-Koura refuses to let AI agents run in his production environment. As CISO at ruby, the parent company of Ashley Madison, he's protecting data where a breach doesn't just expose PII but reveals people's most private thoughts and relationships across a global user base. George tells Jean why the hardest data security challenge is still foundational: too many leaders in the space can't distinguish structured from unstructured data, and organizations keep throwing agents at the problem without understanding the manual processes they're trying to automate, which is exactly why they're not seeing ROI on their AI spend.
George is also pitching the Canadian federal government on a concept he calls the AI Training Data Bill of Material (TDI BOM), modeled after SBOMs: a compliance process that produces a verifiable report ensuring the provenance of data used to train models. He cites studies showing that corrupting less than half of 1% of a model's training data can compromise the entire model, and if that model runs targeting data for defense systems or critical infrastructure like water treatment, the failure mode goes well past data loss. He's pushing for TDI BOMs to be required in government procurement, starting with critical infrastructure supply chains, as a step toward digital and data sovereignty. On the commercial side, George co-founded Very Data Free, a veteran-founded secure-by-design platform he describes as "eBay for your data," built to let organizations sell or loan proprietary datasets for AI model training. The conversation also covers how the SIEM-era centralized security model was built for log aggregation and breaks down at petabyte-scale file data, and why GenAI is forcing organizations to finally secure unstructured data environments they've been ignoring.
Topics discussed: Refusing to let AI agents access production logs and environments
AI Training Data Bill of Material as a government procurement requirement
Model poisoning risks at sub-0.5% training data corruption thresholds
Mapping manual processes before AI automation to prove ROI
Centralized SIEM-era architecture failing at petabyte-scale unstructured data
GenAI forcing organizations to secure previously ignored file environments
AI-generated fake passports and government IDs bypassing identity verification
Hiring self-taught operators over certification-heavy candidates for SOC teams
More from Future of Data Security
When agentic AI systems make multiple tool calls, store data, and act autonomously, static compliance frameworks fall apart. Kevin Bauer, CISO at Chainalysis Government Solutions, argues security teams need continuous monitoring of which controls were in effect at runtime and whether those controls actually worked. Kevin led the first blockchain analytics product to FedRAMP Moderate authorization, where data sensitivity emerged through attribution to real-world entities. He draws direct parallels between those boundary-setting challenges and today's governance gaps in agentic AI, where data classification must drive compliance obligations in real time.
The gap between an exploit's first use and its public disclosure has gone negative. Agentically deployed attacks now reach their target before the vulnerability behind them is ever publicly reported, a shift most security programs are still catching up to. Tom Doughty, Chief Information Security Officer at Generate:Biomedicines, walks Jean through what that shift demands from a security team, from agentic countermeasures to a different approach to non-human identity, plus what changed when he moved a 25-year financial services security career into an AI-native biotech (generatebiomedicines.com) built on machine-generated molecules.
Gopuff runs 200-plus AI agents and 100-plus MCP servers across three public clouds, and Ravi Nori is the one securing all of it. The Head of Cyber Security at Gopuff brought a defense-industry classification ethos to a cloud-native retail platform where customer data sits at the top of the priority chain. Ravi walks Jean through Gopuff's three AI deployments, the shadow AI sprawl he uncovered during an environment scan, and why agent security now spans three surfaces: desktop, browser, and cloud. He also breaks down how boards push for AI adoption while security lags behind.
What if the file itself decided who gets access, carried its own encryption, and refused unauthorized requests? Georgeo X. Pulikkathara, Global CIO and CISO at iMerit Technology, sees that future arriving fast. With 34 years in IT security and active service as a Colonel in the US Army Reserve, Georgeo breaks down 40-minute AI-enabled breaches, the flood of non-human accounts, and why human oversight remains the one control that will never go away.
Gartner predicts 40% of enterprise applications will incorporate agentic AI by year's end, but Jon Rohrich is asking a more basic question: why? Leading Avanade's Canadian cybersecurity practice with nearly 30 consultants, Jon keeps watching proof of concepts succeed, only to collapse when unmodeled risks surface in production.