OpenAI’s Dave Aitel talks Aardvark, economics of bug-hunting with LLMs

(Presented by ThreatLocker (https://threatlocker.com/threebuddyproblem): Allow what you need. Block everything else by default, including ransomware and rogue code.) Three Buddy Problem - Episode 77: New React2Shell data from Microsoft, fresh Apple and Cisco zero-days already in the wild, and state-linked campaigns from Russia and China that show a merging of espionage, crime, and infrastructure disruption. Plus, the US government's push to enlist private firms in offensive hacking, letters of marque for cartels, new discovery of spyware used against journalists in Belarus, and Amazon catching North Koreans via keystroke latency. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

(Presented by ThreatLocker (https://threatlocker.com/threebuddyproblem): Allow what you need. Block everything else by default, including ransomware and rogue code.) Three Buddy Problem - Episode 76: On the show this week, Costin walks through how a single Romanian documentary kick-started nationwide protests, exposing how corruption can be perfectly legal when the law itself is gamed, and why this moment feels different, darker, and more consequential than past flare-ups. Plus, news on the React-to-Shell exploitation wave overwhelming the internet, why patching is structurally hard, and how APTs and criminals are converging on the same fragile dependency chain. Along the way, they take aim at Microsoft’s shrinking transparency, the limits of vendor trust, and what it really means when defenders are told (again) to just patch and pray. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

(Presented by ThreatLocker (https://threatlocker.com/threebuddyproblem): Allow what you need. Block everything else by default, including ransomware and rogue code.) Three Buddy Problem - Episode 75: We dig into a CVSS 10/10 unauthenticated RCE bug causing chaos across the internet and early signs that Chinese APTs are already launching exploits, the cascading patch chaos, and a long tail of malware intrusions to come. Plus, commentary on Chrome’s telemetry collection, Microsoft and the "SFI success story," newest BRICKSTORM backdoor intrusions, the US national security strategy, Anthropic's AI popping smart-contract bugs, a secret FBI ransomware-hunting unit getting weird, and a pair of sad stories in the security community. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

(Presented by Material Security (https://material.security): We protect your company’s most valuable materials -- the emails, files, and accounts that live in your Google Workspace and Microsoft 365 cloud offices.) Three Buddy Problem - Episode 74: We attempt to parse the rumor-fog around Microsoft’s CISO at CYBERWARCON and what it reveals about the company’s shifting posture on intel sharing, regulation, and its outsized grip on the security ecosystem. Plus, coverage of the Shai-Hulud npm supply-chain mess, CISA’s mobile spyware guidance, NSO’s legal contortions, a sharp new GRU-linked intrusion from Arctic Wolf. We also discuss the FCC retreating on telco security rules, and the emerging AI arms race shaping how cloud giants hunt threats and how Washington misunderstands all of it. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

(Presented by Material Security (https://material.security): We protect your company’s most valuable materials -- the emails, files, and accounts that live in your Google Workspace and Microsoft 365 cloud offices.) Three Buddy Problem - Episode 73: The buddies react to Google’s release of Gemini 3 and its early performance, new Chrome interface changes landing on users’ machines, and major highlights from CYBERWARCON. We revisit the long-running debate over APT naming conventions, examine Amazon’s latest threat-intel reporting on Iranian activity, and walk through the Cloudflare outage that briefly knocked chunks of the internet offline. Plus, new APT reports from ESET, Positive Technologies, and SecurityScorecard, and China's CN-CERT (now validated claim) that the U.S. government seized billions in Bitcoin tied to the Lubian mining-pool hack. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

Presented by Material Security (https://material.security): We protect your company’s most valuable materials -- the emails, files, and accounts that live in your Google Workspace and Microsoft 365 cloud offices. Three Buddy Problem - Episode 72: We unpack Anthropic’s conflicting self-promotion around the “first AI-orchestrated cyberattack” using Claude Code and the future of automated APT attacks. Plus, Chinese cyber vendor KnownSec falls victim to data breach, fresh accusations that the U.S. stole billions in Bitcoin, Amazon warning about Cisco/Citrix zero-days, Google’s new Private AI Compute and Microsoft kernel zero-day marked as "actively exploited." Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

Presented by Material Security (https://material.security): We protect your company’s most valuable materials -- the emails, files, and accounts that live in your Google Workspace and Microsoft 365 cloud offices. Three Buddy Problem - Episode 71: The buddies travel to Canada for a live recording at the Countermeasure conference, discussing the Google v FFmpeg open-source patching brouhana, ransomware negotiators charged and linked to ransomware attacks, the looming TP-Link ban in the U.S., and the discovery of LANDFALL, an APT attack caught using a Samsung mobile zero-day. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

Three Buddy Problem - Episode 70: Dave Aitel from OpenAI's technical staff joins the buddies to discuss the just-launched Aardvark, OpenAI’s agentic “security researcher” that claims to read code, finds bugs, validates exploits, and ships patches. We press him on where LLMs beat fuzzers, privacy boundaries, human-in-the-loop realities, SDLC budgets, pen-test cadence, and the zero-day economy. Plus, L3 Harris/Trenchant exec pleads guilty to selling exploits to Russian brokers, Kaspersky catches the return of HackingTeam using Chrome zero-day exploit chain, and news of a proposed law in Russia to force researchers to report vulnerabilities first to goverment agencies. Cast: Dave Aitel (https://www.linkedin.com/in/daveaitel/) (Technical Staff, OpenAI), Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

Three Buddy Problem - Episode 69: We dig into news that Apple's iOS 26 has quietly killed the shutdown.log forensic artifact used to spot signs of infections and what it means for threat hunters. Plus, whispers of a million-dollar WhatsApp zero-click exploit that never materialized at Pwn2Own, a surreal court case linking a Trenchant exploit developer to Russian buyers, and Chinese threat intel reports pointing fingers at the NSA. We also discuss calls for the US government to build a structured, lawful ecosystem for private-sector offensive operations to address existing chaos and market gaps. Cast: Juan Andres Guerrero-Saade (https://twitter.com/juanandres_gs), Ryan Naraine (https://twitter.com/ryanaraine) and Costin Raiu (https://twitter.com/craiu).

Three Buddy Problem (Episode 68): The buddies are trapped in timezone hell with cross-country travel this week. In this special episode, we present Juan Andres Guerrero-Saade's LABScon 2025 keynote-day presentation on the state of cybersecurity and why this phase of our collective project has failed, and how to build something smarter, more sustainable, and deeply interconnected in its place. Juanito traces the field’s evolution from chaos to consolidation, weaving in cybernetics, standardization, and the dawning coexistence of human and artificial evaluative power. The result is part philosophical sermon, part rallying cry, an invitation to reject the industry’s slave morality, rethink our tools, and steer the next era of defense with intention.