Do You Know What Your Agent is Doing at 3AM? with Amir Ofek

Your dashboard says your detections are deployed. But are they still detecting? A quiet SOC can mean your defenses are working, or that the data feeding them has disappeared. In this episode, Ron Eddings sits down with Tom Findling, Co-Founder and CEO of Conifers AI, and Robert Pace, VP of Information Security and CISO at Invitation Homes, to confront the blind spots hiding behind green dashboards. Tom unpacks findings from Conifers AI’s research, The Detection Blind Spot, which examined approximately 15,000 detections and found that 47% needed attention. He describes how telemetry gaps, shifting environments, and detection drift can leave teams believing they are protected when the underlying controls are no longer operational. One striking example: an attacker gradually reduced log-reporting servers from roughly 10,000 to just 10 without stopping logging altogether. Robert challenges security leaders to move past red-yellow-green reporting and the habit of focusing only on critical and high-severity findings. Together, they explore continuous detection validation, AI-assisted tuning and runbook maintenance, the value of understanding the business, and why humans must remain accountable as agentic SOC tools take on more work. They also look ahead to the rise of the “threat operator,” a security practitioner able to connect intelligence, detection, investigation, and response. Impactful Moments 00:00 - Introduction 01:53 - The Detection Myth 03:02 - Silence Is Not a Signal 03:24 - Meet Tom & Robert 6:33 - The 47% Blind Spot 08:10 - A CISO Gets Vulnerable 10:13 - The 10,000-to-10 Attack 13:06 - AI for Continuous Checking 16:09 - Cutting 70% of SOC Alerts 17:28 - Learning the Business 21:36 - CISO-Vendor Relationships 23:21 - Four Criteria for New Tech 25:46 - The "Sorry" Moment 27:24 - The Future: Threat Operator 29:03 - Who Gets the Raise 30:47 - What's Watching the AI? 31:28 - Homework Assignment for Viewers Links Connect with our guest, Tom Findling, on LinkedIn: https://www.linkedin.com/in/tomfindling/ Learn more about Conifers AI: https://conifers.ai Connect with our guest, Robert Pace, on LinkedIn: https://www.linkedin.com/in/robert-pace097/ ----- Check out our upcoming events: https://www.hackervalley.com/livestreams Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com Become a sponsor of the show: https://hackervalley.com/work-with-us/

What would make a government switch off its own websites on purpose? And how does a flaw in the software behind most of the web stay hidden for 14 years? In this solo episode, Ron opens the case file on Black October, the month PHP2Shell pushed governments, hospitals, and banks offline around the world. Ron starts with a rewind to Heartbleed in 2014 and the “change your password” scramble that followed. Then he explains what PHP2Shell is, why AI was the key to finding it, and how three separate teams found the same secret within seven weeks. He also walks through the timeline that doesn’t add up, including someone exploiting the flaw four days before anyone was warned. Then he zooms out to the rest of the month, from supermarket freezers in France turning into ovens to a $6 billion crypto heist that may have started with one engineer’s laptop. Ron closes with three habits that protect you even if you’ve never written a line of code: one password per site, multi-factor authentication, and keeping your devices updated. Impactful Moments 00:00 - Introduction 01:23 - Rewinding to Heartbleed in 2014 04:11 - Chapter 1: The Day The Lights Went Out 05:43 - Why going dark was the smart move 06:44 - Chapter 2: What Scared Them 07:55 - How AI chained the exploit together 08:24 - A 14-year-old door hiding in plain sight 08:36 - Chapter 3: The Four-Day Mystery 09:59 - Three teams, one secret, seven weeks 12:02 - Ron’s hot take on who knocked first 13:59 - Chapter 4: Black October 14:25 - Freezers turned ovens and a $6B heist 17:44 - Chapter 5: What This Means For You 19:25 - Three habits that actually protect you 22:24 - From years to weeks to days Links Check out our upcoming events: https://www.hackervalley.com/livestreams Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com Become a sponsor of the show: https://hackervalley.com/work-with-us/

If someone wiped your cloud environment tomorrow, how long until you’re back in business? Not what the plan says. What would actually happen? Ron sits down with Jake Bernardes, CISO at Gambit Security, who got his first CISO job by sneaking into a company’s office and sitting in the CEO’s chair. He has since run security for tech companies and early-stage security vendors, and he hosts the Random Access Memories podcast. Jake shares the social engineering tricks that got him past security guards. He also explains why voice phishing now takes almost no effort. Then he walks through Gambit’s research on an AI-run attack that cost about $25 per target and needed fewer than ten non-technical prompts. Along the way, he takes on phishing simulations, AI guardrails and cyber insurance. Then comes recovery, the part of security Jake says is still broken. His take: tabletop exercises are “Dungeons & Dragons for executives.” Instead, figure out your minimum viable business, the systems you can’t live without, and prove you can bring them back. Ron closes by tying it to the grocery chain outage that hit his family’s workplace. Impactful Moments  00:00 - Introduction 03:20 - Busting the myth: the cloud has your back 05:45 - Seven minutes of agentic ransomware 10:35 - Breaking into a CEO’s office for a CISO job 12:15 - Paying strangers to get past security guards 15:25 - People don’t hack in, they log in 18:55 - The ASOS incident and the short-stock theory 20:00 - Too much exposed to the public internet 23:05 - Anatomy of a $25 AI attack 27:10 - Translating the attacker’s Chinese prompt 29:40 - Can CISOs rely on AI guardrails? 32:25 - Why cyber insurance is a tick box 35:30 - Minimum viable business for 2027 38:25 - Ron digs into the Giant Food outage Links Connect with Jake Bernardes on LinkedIn: https://www.linkedin.com/in/jakeleobernardes/ – Check out our upcoming events: https://www.hackervalley.com/livestreams Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com Become a sponsor of the show: https://hackervalley.com/work-with-us/

What would you do if your phone rang and the caller ID showed your own company? Would you pick up? Would you help the person on the other end? This Cybersecurity Awareness Month, Ron goes solo to break down a breach that started with that exact call. Attackers spoofed Astrana Health’s main corporate phone number, posed as coworkers, and talked their way into systems connected to roughly 1.5 million patients. Ron starts with a rewind from phone books to data brokers, including a live search on a people-search site that turned up his own family. From there he walks through the SEC filing, sends a swarm of AI research agents after the public record, and lines up the suspects behind the attack. The uncomfortable part is that we spent a decade training people to spot a bad email while attackers moved to the phone. Voice phishing is now the second most common way attackers get in, and resetting credentials doesn’t fix the instinct to help when a familiar number calls. Ron closes with three practical ways to stop making it easy for them, starting with one simple habit: hang up and call back. Impactful Moments 00:00 - Introduction 02:25 - The rewind: from phone books to data brokers 03:15 - What a people-search site found on Ron’s family 05:30 - A breach that started with a phone call 06:15 - Inside Astrana Health’s SEC filing 07:55 - Why spoofing the main number changes everything 09:05 - How the attack likely played out 11:35 - Technical fixes for a human attack 12:50 - The time Ron got got 13:35 - Why attackers want healthcare data 15:15 - 1.5 million patients’ worth of exposure 17:15 - Voice phishing climbs to #2 in M-Trends 19:35 - Three ways not to make it easy for them 21:05 - How to run a phone drill Links Check out our upcoming events: https://www.hackervalley.com/livestreams Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com Become a sponsor of the show: https://hackervalley.com/work-with-us/

What happens when your marketing team, your sales team, and the person down the hall can all ship a full app from a single prompt? And what is that app quietly connected to? In this episode, Ron sits down with Roi Nisimi, Principal Security Researcher, and Yonatan Levi, AI Security Researcher, of Orca Security. They built apps on the most popular AppGen platforms and then tried to break into them. Ron, Roi, and Yonatan get into what's really running behind a prompt-built app, and what Orca's team found when they went looking for trouble. In one case, a single misconfigured app gave them a way into every integration connected to the workspace. They also explain why Supabase has become a favorite of builders and attackers alike, and share the security questions they ask before sending a prompt. The bigger tension is speed. With no code review between the prompt and production, Roi argues that velocity has broken the threat models that security relied on for years. And when the agent makes every decision, nobody fully understands what was built, not even the security researchers. Impactful Moments 00:00 - Introduction 02:15 - Busting the vibe coding myth 04:40 - Inside Orca's State of AI report 05:55 - The unpatched AI SDK problem 07:40 - What is AppGen? 10:20 - Who builds the backend? 12:25 - Inside Orca's AppGen hackathon 14:10 - Thinking like a citizen developer 15:15 - Who's really building these apps 19:25 - One misconfigured app, every integration exposed 20:45 - Why Supabase is the holy grail 23:00 - Checking your own exposure 24:30 - Why the SDLC is breaking down 26:55 - Prompting for a secure app 28:15 - Don't hand the agent every decision 31:20 - Staying sharp by never stopping learning 33:35 - From protecting environments to protecting apps Links Connect with Roi Nisimi on LinkedIn: https://www.linkedin.com/in/roinisimi/  Connect with Yonatan Levi on LinkedIn: https://www.linkedin.com/in/yonatan-levi-b22168258/  Register for Orca’s virtual Builder Exchange: https://orca.security/builder-exchange/?utm_source=HVN&utm_medium=3rd-part Read the security findings in Orca’s 2026 State of AI Security Report: https://orca.security/lp/2026-state-of-ai-security-report/?utm_source=HVN&utm_medium=3rd-party  – Check out our upcoming events: https://www.hackervalley.com/livestreams  Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com  Become a sponsor of the show: https://hackervalley.com/work-with-us/ 

Your newest hire has a Google Workspace account, a Slack login and a laptop. It never sleeps, and it never asks before it acts. So who owns its identity? Ron welcomes back Ashish Rajan, CISO at Techriot.io and researcher with AI Security Lab, who spent 17 years in identity, cloud security and security leadership. His take: vendors haven't solved AI identity, and the open questions are on your side, not the product's. Ashish got into identity after failing his OSCP three times: if he couldn't break in, he wouldn't let anyone else in. Now he reviews Ron's own setup, where Hacker Valley's AI agent, Jennifer Romani, has her own accounts just like an employee. Ashish explains how that call changes from a one-laptop small business to an enterprise where every developer's machine acts like five. They also cover the Gemini breach headlines, swarms of agents and AI Security Lab, which Ashish describes as what OWASP is for AppSec, but for AI security. The uncomfortable truth: defenders have the same models attackers do, but attackers don't wait for a pull request review. Ashish calls that a people problem, not a technology problem. He also shares the question he asks before handing anything to an AI agent: can this be reversed? Impactful Moments 00:00 - Introduction 02:30 - Busting a myth: AI identity isn't solved 05:30 - Three OSCP attempts and a pivot to IAM 07:00 - Why AI identity is more complex than ever 09:15 - Hot or not: giving an AI agent its own accounts 11:10 - One laptop, five machines in the enterprise 12:50 - AI gateways and developer observability 14:05 - Where Ashish would spend his AI security budget 18:20 - Why defenders don't hack themselves first 19:05 - The real problem is people 22:15 - Swarms of agents are coming 23:50 - What AI Security Lab is building 26:35 - Never share your credit card with AI 28:40 - Ron's callback: can it be reversed? Links Connect with Ashish Rajan on LinkedIn: https://www.linkedin.com/in/ashishrajan/ Listen to Ashish's first Hacker Valley Studio episode: https://www.podbean.com/pw/pbblog-bpaij-51b280 –  Check out our upcoming events: https://www.hackervalley.com/livestreams  Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com  Become a sponsor of the show: https://hackervalley.com/work-with-us/ 

What if someone got full admin access to your company's platform, including your CRM, your payments app, and your private messages, and the only tool they used was an AI chatbot? Can AI models really find zero-days on their own, or is that just the headline? And if AI can do the attacker's job, who's actually holding the scissors? In this solo episode, Ron Eddings shares his own methodology for offensive assessments with AI. He talks about the models he trusts, the tools he uses to get agents working together, and a real assessment where his agents turned SQL read access into admin control and unlocked the API keys stored inside. Ron also calls out what's hype and what's real with today's models, and why the agent still needs a skilled person behind it. For defenders, Ron covers what attackers go after once they're in, and why the fundamentals like asset inventory are still where every strong security program starts.  He closes with the bigger picture: anyone, good or bad, now has powerful intelligence on hand, and it's on all of us to look out for each other. Impactful Moments 00:00 - Introduction 02:25 - Can AI Really Find Zero-Days Alone? 04:20 - The Real Danger Is Human Intent 05:00 - Why Grok 4.6 Tops Ron's List 07:55 - Ron's Three-Model Assessment Workflow 09:50 - Maestro, Interceptor, and Agent Ensembles 11:20 - Privilege Escalation and Persistence With LLMs 12:50 - Why AI Hacking Feels Like Cheating 14:20 - Ron Called Automated Security in 2015 16:05 - The Lazy Way to Hack 17:55 - From SQL Read Access to Admin 21:05 - What Attackers Want After Getting In 23:40 - Asset Inventory Is Security Flossing 27:05 - Why AI Is Like Scissors 29:05 - Waymos, Robotaxis, and Physical AI Risk Links Check out our upcoming events: https://www.hackervalley.com/livestreams  Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com  Become a sponsor of the show: https://hackervalley.com/work-with-us/ 

Every AI agent in your environment inherits someone's permissions, and most teams have no idea what those agents are actually doing with them. Amir Ofek, CEO and co-founder of Aizome, shares how to make every agent accountable before the damage shows up as a $150,000 bill. In this episode, Amir explains why human and machine identity tools both fail for AI, why agents act more like eager, naive interns than employees, and how a new category called ARISE (Agentic Runtime Identity Security Enforcement) is emerging to fix it.  This one's for anyone building with AI agents right now (so, most of us). Amir's take: the biggest danger isn't a breach. It's agents quietly doing the wrong thing at scale while nobody's watching. Impactful Moments 00:00 - Introduction 02:30 - Busting the Myth: "I Know What My Agents Are Doing" 05:30 - The Samurai Story Behind Aizome's Name 08:25 - Why Not All AI Agents Are Born Equal 11:15 - Where Enterprises Are Actually Deploying AI Agents Today 14:25 - What Claude Cowork Inherits Without You Realizing It 17:15 - Machine Identity vs. Human Identity vs. AI Identity 19:35 - Treating AI Agents Like Eager, Naive Interns 23:00 - The Biggest AI Risk Isn't Security, It's Cost 25:45 - Meet Ito: Aizome's Supervisor Agent 26:40 - Inside the New ARISE Category 33:35 - What Aizome Actually Does 35:30 - The Callback: Was the Myth Wrong, or Dangerously Wrong? Links Connect with Amir Ofek on LinkedIn: https://www.linkedin.com/in/amirofek/ Learn more about Aizome: https://www.aizome.ai/  –  Check out our upcoming events: https://www.hackervalley.com/livestreams  Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com  Become a sponsor of the show: https://hackervalley.com/work-with-us/ 

Cybersecurity has survived cloud, mobile, and a dozen other "biggest disruptions of our lifetime," and each one felt unprecedented in the moment. In this episode, Ron sits down with Alyssa "Dr. Jay" Abdullah, Deputy CISO at MasterCard, who started her career as a radio DJ and has since worked inside the White House, Lockheed Martin, Stryker, and Xerox before landing in payments security. Ron and Dr. Jay trace her path from spinning records to securing global payment infrastructure, and dig into why she'd argue cloud, not AI, was the real turning point in her career. They cover what convergence actually looks like when AI, cloud, and synthetic identity start overlapping, why curiosity matters more than fast answers, and what it means when employees start bringing their own trained AI agents to work. The conversation closes on something most teams haven't fully reckoned with yet: AI agents that carry their own identity, independent of the humans who built them, and what that shift demands from the people responsible for securing them. Impactful Moments 00:00 - Introduction 02:25 - Busting the AI hype myth 04:40 - From radio DJ to Dr. Jay 06:10 - A day in the life as Deputy CISO at Mastercard 07:25 - Why AI hasn't disrupted her world 08:50 - White House tech through the decades 12:30 - Smartphones, wearables, and what's next 13:30 - Defining convergence for 2026 14:50 - When AI meets quantum computing 17:20 - Bring your own AI agent to work 19:00 - Negotiating salary in tokens 21:05 - Teaching curiosity over answers 23:50 - Overhype equals overtrust 27:10 - The future of tier one and autonomous SOC 29:50 - Hot take: AI in the SOC 31:35 - Predictions: AI identities and the human outside the loop Links Connect with Alissa "Dr. Jay" Abdullah on LinkedIn: https://www.linkedin.com/in/dralissajay/  –  Check out our upcoming events: https://www.hackervalley.com/livestreams    Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com    Become a sponsor of the show: https://hackervalley.com/work-with-us/ 

Security teams spent decades begging for more logs. Now the enterprise is generating petabytes a day, and the thing drowning in it isn't just the SOC anymore, it's your AI agents too. In this episode, Ron sits down with Myke Lyons, CISO at Cribl, who cut his teeth in telemetry and logging decades ago and has landed right back there in the age of AI. Ron and Myke dig into why most telemetry failures aren't data problems at all, they're decisions nobody made about why the logs are being collected in the first place. Myke breaks down what to track on every agent in your environment, why token spend belongs on the security team's plate, why dashboards are quietly dying, and why treating an AI agent like just another employee is a mistake that's going to bite security teams hard. Underneath it all is one question Myke keeps circling back to: do you actually know what normal looks like for every agent in your environment? Impactful Moments 00:00 - Introduction 01:50 - Myth Busting: AI Agents Aren't Just Another User Account 04:25 - Meet Myke Lyons, CISO at Cribl 05:05 - What it means to run security at a telemetry company 06:10 - From gigabytes of logs at GE to petabytes today 07:45 - MITRE ATT&CK, Cribl's new APEX framework, and orienting telemetry 10:20 - Why most telemetry problems are decision problems, not data problems 13:20 - OCSF and how security teams are rethinking their schemas 14:25 - Why the dashboard is dying 17:35 - What Myke wants to track about every AI agent 20:10 - How to spot an agent going rogue 23:15 - Making your data AI-ready and the case for schematizing everything 27:10 - Tokenomics: treating AI spend as a security responsibility 29:05 - The non-negotiable logs every org should be collecting 31:50 - Hot takes: build vs. buy, tier two to three, and Myke's daily AI briefing 33:35 - Closing thoughts and outro Links Connect with Myke Lyons on LinkedIn: https://www.linkedin.com/in/mykelyons/ Learn more about Cribl: https://cribl.io/  –  Check out our upcoming events: https://www.hackervalley.com/livestreams   Love Hacker Valley Studio? Pick up some swag: https://store.hackervalley.com   Become a sponsor of the show: https://hackervalley.com/work-with-us/