4 Things IT Leaders Should Do as AI Reshapes Cybersecurity

CYBER BYTES | AI CYBERSECURITY

In April 2026, Anthropic launched Project Mythos, its most capable AI model, which autonomously discovered over 23,000 vulnerabilities across 1,000+ open-source projects — including bugs that had gone undetected for over 25 years. Anthropic responded by creating Project Glasswing, bringing together Microsoft, Google, Apple, CrowdStrike, and dozens of other tech giants to scan and patch critical code before threat actors get access to the same power. Within weeks, Microsoft launched MDASH and OpenAI launched Daybreak, making it clear the AI-driven cybersecurity race is fully underway.

In this episode of Cyber Bytes, Alex Ryals breaks down what happened and the four things every IT leader needs to act on now:

• Why your patch window just shrank from weeks to hours
• How AI-powered defense has to match AI-powered offense
• What “shift left” means for your security and development teams
• Why AI governance and risk management are now the same conversation