21 mins S5 #14 Nov 18, 26 Nvidia, Fireworks AI, and the Push to Block Frontier AI Nvidia’s CEO exits, Fireworks AI is breached twice, and nonprofits sue to block frontier AI development. Martin Reynolds and David Cannon dig into vendor lock-in, supply chain risk, and why the tried-and-tested engineering practices still matter when you build with AI. Stop talking. Start shipping. ShipTalk is brought to you by Harness.
18 mins S5 #13 Nov 11, 26 Why the Vulnerability Tsunami Has Already Started October brought a flood of AI-driven vulnerabilities, and it may only be the start. Martin and Adam break down the "vulnerability tsunami" and what teams can do about it. In this episode: • Black October: why open-weight models catching up to frontier cyber models changed everything, and who benefits when AI companies sell the fire extinguishers • Vuln Zero: Anthropic's Jason Clinton predicts that within 18 months there will be zero exploitable vulnerabilities left to find. Is that a real fix, or just a destination? • Why 98.5% of CVEs since 2018 have never been exploited, and why prioritization and fast remediation may matter more than chasing zero • The FBI and CISA warning about attacks on internet-exposed refrigeration controllers, after hits on French supermarkets and blood storage sites • A red team story where an AI agent swarm kept going after it was "turned off" • Why vulnerability management is a security problem, an engineering problem, and a business problem • Plus: MS Paint got hacked ShipTalk is brought to you by Harness. We break down how software delivery is actually changing in the age of AI. Stop talking. Start shipping.
25 mins S5 #12 Nov 04, 26 Google's Stock Drop Reveals What AI Is Missing Is the AI bubble starting to burst, or is the industry just beginning to question the cost of artificial intelligence? Google's parent company, Alphabet, reported $51.4 billion in quarterly capital expenditures, raising questions about AI spending, profitability, and when investors will see meaningful returns. Meanwhile, Anthropic has filed for its IPO, revealing risks that have cybersecurity leaders questioning the future of the AI industry. And at GitHub Universe 2026, the explosion of AI-generated code is raising another concern: are developers shipping better software, or just more of it? In this episode of ShipTalk, hosts Adam Arellano and Martin Reynolds are joined by Jake Bernardes, CSO at Gambit, to unpack the biggest developments in AI, cybersecurity, and software delivery. They debate whether we're entering an AI bubble correction, the risks behind Anthropic's IPO, why traditional shift-left security may be failing in the age of AI coding agents, and whether enterprises are measuring AI productivity all wrong. The conversation also explores software supply chain security, enterprise AI ROI, and the growing risks of giving developers and AI agents access to critical infrastructure. **ShipTalk is brought to you by Harness.** Real conversations about what's changing in AI, DevOps, cybersecurity, and software delivery.
23 mins S5 #11 Oct 28, 26 How AI Is Changing Cybersecurity and Exposing Decades of Bad Code Critical PHP vulnerabilities discovered with AI. A major LiteLLM software supply chain attack. Unauthorized AI agents accessing US bank accounts. October 2026 delivered a remarkable week of cybersecurity news—but is AI creating new security threats or exposing weaknesses that have existed for decades? Martin Reynolds and Adam Arellano break down the PHP2Shell vulnerabilities, the Chainguard patch controversy, AI agent security risks, and the growing tension between software engineering and cybersecurity teams. As AI accelerates vulnerability discovery and raises new questions about software security, the hosts debate who should be responsible for securing the next generation of applications—and whether this is just the beginning of a new era in cybersecurity. ShipTalk — https://www.shiptalk.io Harness — https://www.harness.io Follow Adam Arellano: https://www.linkedin.com/in/adamrossarellano/ Martin Reynolds: https://www.linkedin.com/in/martinreynolds/ Brought to you by Harness.
22 mins S5 #10 Oct 21, 26 OpenAI’s Hidden Mic Scandal, Microsoft’s 1,400 Patches & Japan’s Cyber Emergency An OpenAI engineer is arrested after a hidden recording device is discovered in an executive conference room. Who was he recording, and who was the information intended for? Microsoft releases a record-breaking 1,400+ security fixes in a single Patch Tuesday, including three zero-day vulnerabilities. Is AI helping discover security flaws faster than companies can fix them? Meanwhile, Japan declares a cybersecurity emergency as AI threatens to make large-scale attacks easier than ever. And the ShinyHunters hacking group claims to have stolen sensitive FBI personnel records, raising serious national security concerns. Martin Reynolds and Adam Arellano break down the latest cybersecurity headlines, the growing risks of AI-powered attacks, and whether defenders can keep up. ShipTalk is brought to you by Harness. ShipTalk, AI security, cybersecurity, Five Eyes, AI national security, Fable, Mythos, security workforce, ISSA, software delivery, DevOps, AI jobs, CISO. Adam Arellano: https://www.linkedin.com/in/adamrossarellano/ Martin Reynolds: https://www.linkedin.com/in/martinreynolds/