Frontier AI models are increasingly used by both companies to identify system vulnerabilities and by malicious actors to exploit them. Recent incidents involving the PHP programming language, which supports up to 70% of global websites, demonstrate this shift. Security researchers utilized OpenAI’s GPT-5.6 Sol to find critical bugs, while attackers began targeting these vulnerabilities before patches were released, leading to millions of exploit attempts.
The cybersecurity community reports a significant rise in threats, with Microsoft documenting 1,453 new vulnerabilities in October 2026 alone. Industry leaders, including Dell Technologies, are responding by integrating frontier models into defense strategies, such as participation in Anthropic’s Project Glasswing. Experts emphasize that the barrier to entry for sophisticated attacks has lowered, necessitating a shift toward robust recovery and resilience postures.
Data protection is becoming a foundational control layer for enterprise AI, as the reliability of AI systems depends on the integrity of data pipelines and recovery processes. Companies like Infinidat are focusing on immutable snapshots and cyber detection to ensure clean recovery points. Furthermore, the threat of quantum computing looms, with a DigiCert study showing that only 7% of organizations have deployed quantum-safe cryptography.
Research using Anthropic’s Claude Mythos Preview has revealed improved methods for attacking cryptographic algorithms, including the weakening of the HAWK signature candidate. Experts warn that attackers no longer require quantum computers to make progress, as classical compute can now be used to exploit these weaknesses. Consequently, infrastructure-level security and proactive data recovery have become critical for enterprise compliance.
