As noted by Infosecurity Magazine, the median number of high or critical vulnerabilities in cloud workloads increased fourfold between May and November 2026, a trend attributed to the rapid integration of frontier artificial intelligence.
Datadog, a cloud observability and security firm, released its State of Cloud Security 2026 report on Nov. 10. The findings indicate that high-severity flaws surged from four in May to 16 in November. These vulnerabilities often stem from third-party libraries within software applications. Beyond software flaws, the report highlights risks associated with long-lived credentials, particularly in AI inference services and continuous integration and deployment pipelines. Nearly 50% of organizations using Gemini utilize long-lived API keys, while 30% of users maintain access keys older than three years. Furthermore, supply-chain attacks are increasingly automated. Datadog observed that 60% of organizations adopt new dependency versions within 24 hours of release, often without sufficient vetting. With an average of 149 dependencies per cloud workload, the potential for compromise is significant. Experts recommend implementing dependency cooldown periods to allow time for the identification of malicious libraries and transitioning to short-lived identity credentials to minimize the impact of potential breaches.
Source: Infosecurity Magazine
