Full Summary
This Friday, May 22nd, IBM and several other security firms are urgently boosting AI security efforts as advanced AI models accelerate the discovery and exploitation of software vulnerabilities. Both IBM and security firms testing Anthropic's Mythos AI confirm these models are "incredibly good" at finding flaws. IBM has joined Project Glasswing to protect critical software infrastructure and is integrating AI into vulnerability analysis. Simultaneously, Anthropic's Mythos AI has already identified critical Apple macOS flaws, with researchers from Calif using it to find a "first public macOS kernel memory corruption exploit on Apple M5," which could grant full device access. What nobody expected is the sheer speed. The 2026 Threat Intelligence Index reports a 44% year-over-year increase in public-facing application exploitation, with AI directly aiding adversaries. This means that while AI is great at *finding* vulnerabilities, a study found 45% of AI-generated code *contains* security flaws, and AI-generated pull requests have 1.7 times more issues than human-written code. Zscaler is acquiring Symmetry Systems to enhance its security platform, specifically to understand how AI agents, applications, and data interact, addressing a gap where traditional access control falls short. This acquisition aims to integrate data security and AI governance tools into Zscaler's Zero Trust Exchange platform. Check Point is also acquiring Deepchecks to boost confidence and trustworthiness in autonomous security systems, especially as generative AI introduces new operational risks like hallucinations. This all means that the software you use, from your operating system to critical applications, faces an unprecedented level of AI-driven threat and opportunity. Expect more frequent software updates and a higher demand for robust, AI-powered security solutions to protect your data and devices.