Datadog says high and critical cloud flaws quadrupled since May
Original title: Cloud Vulnerabilities Surge Fourfold in Less Than a Year
Among monitored workloads, the median count rose from four in May to 16 in November, Datadog claims.
Datadog's State of Cloud Security 2026, published November 10, says the median number of high or critical flaws in cloud workloads rose from four in May to 14 in June and 16 in November. The figures come from software composition analysis of third-party libraries in applications monitored across thousands of organizations. Datadog links the jump to frontier AI after Anthropic's Mythos release and also flags long-lived AI and CI/CD credentials and rapid dependency adoption.
Unconfirmed: Datadog attributes the fourfold rise to frontier AI; that cause is not independently confirmed.
- Nearly half of organizations using Gemini keep long-lived Gemini API keys.
- Nearly 30% of AWS IAM and Google Cloud service-account users have keys older than three years.
- Sixty percent of analyzed organizations run dependency versions released less than 24 hours earlier.
- Cloud app workloads average 149 dependencies; the average organization has 568.
- Datadog recommends short-lived workload identities and dependency cooldowns.
Impact: More severe flaws, exposed keys, and hijacked dependencies in cloud apps
- 1.Use short-lived IAM roles, managed identities, or service accounts for AI inference instead of static keys.
- 2.Add dependency cooldowns so newly released library versions are not adopted automatically.
- Products
- AWS IAMGoogle Cloud
- Organizations
- Anthropic
From the original article
2 min read at infosecurity-magazine.com
AI is amplifying the impact of vulnerabilities, long-lived credentials and other cloud risks, says Datadog
By Phil Muncaster · infosecurity-magazine.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.