This week on Enterprise Security Weekly, Adriel Desautels joins to talk about the state of penetration testing — why it hasn’t done much to help security teams over the last 20 years, and why AI won’t fix it either.
Then there’s Meta. The company’s aggressive AI reorganization has been making waves, and not in a good way. Engineers inside the company describe a chaotic environment where dogfooding AI tools has created more problems than it solves. The Pragmatic Engineer newsletter published a detailed analysis of what’s going wrong, and it’s relevant to security: heavier AI use at these companies appears to correlate with more availability and security issues.
Meta’s CTO Andrew Bosworth admitted the AI reorg was, in his words, “atrocious.” That’s not exactly reassuring when you’re trying to build secure systems.
The weekly enterprise security roundup also covers an AI SOC vendor shutting down, cybersecurity vendor layoffs, cascading breaches, and the odd detail that criminals don’t trust AI either. Some developers now refuse to code without AI assistance, even when you pay them not to use it. And Midjourney is apparently becoming a healthcare company now.
