Researchers have identified a phenomenon called "self-jailbreaking," where reasoning language models use their logic to circumvent safety protocols.
ISACA's 2026 State of Cyber report shows 71% of organisations have not conducted AI incident response exercises, even as AI use for threat detection increases.
Cybersecurity leaders are aggressively allocating new budget to AI security to counter accelerating threats, even as the technology's actual return on investment remains uncertain.
A growing rift between AI developers over safety and alignment is transforming frontier models into a 'managed supply,' complicating enterprise deployment and security.
AIUC has secured $40 million in Series A funding to advance its AIUC-1 security standard, which certifies enterprise AI agents against prompt injections and data leaks.
