Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
A cloud cost anomaly is an unexpected jump in spend. Often it is a bad deployment—but it is also one of the most reliable early signals of stolen credentials and cryptomining.

What Is a Cloud Cost Anomaly?
A cloud cost anomaly is a deviation in spend that the usual pattern does not explain: a compute line that triples overnight, data transfer charges from a region you do not operate in, or a service appearing on the bill that nobody recognises. Finance teams treat it as a budget problem. Security teams should treat it as a detection opportunity, because for several common attacks the invoice moves before any alert fires.
Stolen cloud credentials are monetised in predictable ways, and most of them cost money.
The pattern is also useful in reverse. Because mining and inference abuse are compute-intensive, they are far harder to hide than the quiet, low-volume activity of an espionage actor. Cost monitoring is a cheap control that catches a noisy class of attacker quickly.
All three major providers offer anomaly detection built on historical spend: AWS Cost Anomaly Detection, Microsoft Cost Management anomaly alerts, and Google Cloud budget alerts with forecast-based thresholds. They are free or close to it, and they are routinely left unconfigured.
Configure them per account, subscription, or project rather than only at the organisation level, because a single compromised development account disappears inside an enterprise invoice. Set thresholds in percentage and absolute terms, so a small account with a tenfold increase still alerts. Route the notification to the security team as well as to finance—an alert that only reaches a monthly billing review has lost most of its value.
Pair the cost signal with service-level detections. GuardDuty raises findings for cryptocurrency-related network activity and anomalous API calls, Microsoft Defender for Cloud produces similar alerts, and both correlate far better with billing data than either does alone. Good cloud logging and monitoring turns a cost alert into an investigation with a timeline.
Work through the question quickly rather than assuming either answer. Which service, which region, and which account? Resources launched in a region you never use are a strong indicator of abuse, because deployment errors usually happen where deployments normally happen. Who created the resources, and through which credential? Check the audit trail for the creating principal and for the events immediately before—a new access key, a new IAM user, a disabled logging configuration, or a changed trust policy around the same time turns a billing question into an incident.
If the answer points to compromise, follow your cloud incident response plan: revoke the credential and its active sessions, quarantine rather than immediately delete the resources so you keep evidence, and review every other account that shared the credential. Deleting the mining instances first is tempting and destroys the forensic record.
Service quotas are an underused control. Lowering default instance limits in accounts that will never need them caps the damage of a compromise and costs nothing. Region restrictions through service control policies or their Azure and Google equivalents do the same. Short-lived credentials remove the long-lived keys that these campaigns depend on, and budgets with hard stops are appropriate for sandbox accounts even if they are not for production.
Finally, agree in advance who pays for an incident like this. Providers sometimes issue goodwill credits for fraudulent usage, but the request is far smoother when you can show the date of detection, the containment action, and the control change that followed.
At Privia Security, we help teams turn billing telemetry into usable detection through our SOC maturation services.
No, and most are not. Forgotten test environments, a misconfigured autoscaling rule, a chatty logging change, or a data migration account for the majority of spikes. The point of treating cost anomalies as a security signal is that the investigation is cheap and the downside of ignoring a genuine compromise is high. Triage by region, service, and creating principal, and most questions resolve in minutes.
Fast. Automated tooling launches high-specification instances across every available region within minutes of a credential being leaked, and published cases describe five-figure charges accumulating within a day or two. That speed is why month-end review is the wrong control and why service quotas, region restrictions, and same-day anomaly alerts matter more than any retrospective report.
Sometimes, as a goodwill credit rather than an entitlement. Your case is strongest when you can show the leaked credential was revoked promptly, the resources were contained, and you have changed the control that allowed it. Open a support case as soon as containment is complete, keep the billing export and audit log extracts, and expect the review to take time.
Use at least two. A percentage deviation from the account’s own baseline catches small accounts that would vanish in an organisation-wide threshold, and an absolute currency threshold catches large accounts where a modest percentage is still a serious sum. Add a separate alert on data transfer out, because egress is the line item most closely tied to bulk data theft.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.