< Back to all clusters

Frontier artificial‑intelligence models are compressing the time needed to discover, craft and launch exploits from weeks to minutes, eroding the traditional advantage defenders once had. Attackers now use automated, prompt‑injection techniques to bypass safeguards in large language models, while security teams are turning the same method against them, forcing malicious AI agents to shut down.

Cisco’s IQ platform, now used by more than 5,000 organizations, offers comprehensive asset visibility to counter these threats. It highlights the growing risk of unsupported, end‑of‑life devices, which accounted for 40 % of the top 100 vulnerabilities last year and are expected to rise in 2026. Recent malware campaigns such as the ClickLock macOS stealer and TELEPUZ/ClickFix chain illustrate the expanding attack surface that blends AI‑driven tactics with classic credential‑stealing methods.