Mythos Preview, a new AI model, developed 181 working exploits for vulnerabilities in Mozilla's Firefox JavaScript engine, achieving register control on 29 more. Mythos Preview's development of 181 working exploits for vulnerabilities in Mozilla's Firefox JavaScript engine, achieving register control on 29 more, demonstrates a powerful new capability in cyber offense, accelerating attack speed and scale.

A significant majority of organizations face costly security breaches due to a lack of cybersecurity skills. Yet, the very AI they adopt for defense now proves capable of sophisticated offensive operations. Eighty-six percent of organizations reported one or more security breaches in the past 12 months, with 52% incurring costs over $1 million, according to Fortinet. For the third consecutive year, 56% of IT leaders cited a lack of cybersecurity skills as a top cause. This persistent human deficit, coupled with AI's new offensive prowess, creates a critical vulnerability for businesses.

Companies must urgently address human skill deficits and rapidly adapt security postures. They need to both leverage and defend against increasingly autonomous, powerful AI-driven threats.

The Persistent Human Gap in a Digital Battlefield

  • Cyber-threat literacy was identified as the top risk in a recent survey, according to Infosecurity Magazine.
  • Shortages in cyber and AI skills were ranked as the third highest risk in the same survey.

These findings reveal that foundational human vulnerabilities, in both understanding and specialized expertise, persist despite technological advancements. This gap means organizations struggle to even deploy advanced security measures, leaving them exposed to both known and emerging threats.

AI's Double-Edged Sword: From Flaw Detection to Exploit Creation

Mythos Preview's 181 successful exploits against Firefox vulnerabilities demonstrate an unprecedented level of autonomous offensive capability. This shifts the advantage to attackers, accelerating attack sophistication beyond what scarce human defenders can match. Such AI-generated exploits frequently bypass traditional detection methods. This rapid evolution of AI suggests the financial burden of cyberattacks will intensify as AI democratizes advanced offensive capabilities for a wider range of threat actors.