LEMON BLOG

Anthropic's Project Glasswing has revealed something important about the future of cybersecurity: AI can now find serious software vulnerabilities at a speed that traditional patching processes may struggle to match.

Apple's macOS security has reportedly faced a major research breakthrough, this time with help from Anthropic's advanced cybersecurity-focused AI model, Claude Mythos. According to reports, security researchers used the AI system to identify vulnerabilities in macOS and assist in building what is being described as the first public macOS kernel memory corruption exploit affecting Apple's M5 silicon.

Artificial intelligence is quickly becoming part of the cybersecurity world, but a recent hacking competition showed that the technology is still far from replacing experienced human defenders and attackers. In a high-stakes cyber defence event, AI agents were placed directly into the action, working alongside and against human teams in a realistic simulation of cyberwarfare.

For the past year or two, a lot of big companies have been "trying out" AI coding assistants the same way they try out any new dev tool: a small pilot here, an optional experiment there, a few enthusiastic teams using it quietly while everyone else sticks to привычный workflows. That kind of cautious rollout made sense. Code is high-stakes: quality, security, maintainability, and team consistency matter more than novelty.

When you see the name Apple Intelligence, you might think about a self-contained "Apple AI". But appearances are shifting. What once looked like an in-house platform is evolving into something more collaborative—an ecosystem of artificial intelligence tools from different developers, brought together under Apple's software umbrella.