Smashing Security podcast #463: This AI company leaked its own code. It’s also built something terrifying

This podcast episode discusses recent cybersecurity issues: Anthropic accidentally published source maps that exposed the Claude Code CLI source (a sensitive intellectual-property/data leak), the company’s internally-developed Mythos model that can automatically find and chain application vulnerabilities into exploits (presenting a high-risk capability), and claims by a hacking group that they accessed Venice’s flood-control pumps and offered root access for sale. The hosts analyze risks around CI/CD and build-server misconfiguration, OT safety implications, and the broader supply-chain and AI-enabled vulnerability-discovery threats.