top of page

What Is Cryptojacking? Signs, Risks, Detection Tools & Prevention

6 hours ago
26 min read
Cryptojacking attack secretly using computer resources to mine cryptocurrency.

Your laptop fan is loud, but you are not doing anything heavy. Your cloud bill jumped, yet nobody launched anything new. In cases like these, the cause can be quiet and boring: someone else is using your computing power to make money. This guide explains cryptojacking in plain words, shows the real warning signs, compares detection tools, and gives safe steps to check, clean up and prevent it.


TL;DR


  • Cryptojacking is the unauthorized use of someone else's device, server or cloud account to mine cryptocurrency.

  • The main harm is stolen computing power: slow devices, higher power use and surprise cloud bills. Data theft happens only if the intruder also did other things.

  • High CPU use alone proves nothing. Confirm with the process, where it came from, how it starts and what it connects to.

  • Tools differ. Task Manager shows load, endpoint tools block threats, and cloud services watch accounts and workloads.

  • Do not just kill a process. Contain the device, keep evidence and close the door the attacker used.


Quicl Answer: What Is Cryptojacking?


Cryptojacking is the unauthorized use of someone else's computing power to mine cryptocurrency. Attackers run mining software on a victim's device, server or cloud account without permission, keep the coins, and leave the victim with slower systems, higher power use and sometimes large cloud bills. It is usually designed to stay hidden.


Illustration of a web application window being inspected with a magnifying glass and a shield, representing a web app security assessment.

Could a weak web app be one open door?


Mining malware can get in several ways, and a weak web app, API or SaaS tool is one of them. A scoped, authorized penetration test checks your own applications for weaknesses an attacker could use, then gives your team a ranked list of fixes. It complements monitoring. It does not replace it.




On this page



What Is Cryptojacking?


Cryptojacking is the secret use of another person's computing power to mine cryptocurrency. Mining is the work of checking and recording crypto transactions. The miner who does this work can earn new coins, but the work needs a lot of processor time and electricity. A cryptojacker skips that cost by making you pay it.


The key word is authorization. If you install mining software on your own computer, that is legitimate mining. If someone else runs it on your device, server or cloud account without your clear permission, that is cryptojacking. The MITRE ATT&CK knowledge base tracks this behavior as Resource Hijacking (T1496) and its Compute Hijacking (T1496.001) sub-technique.


Who Profits and What the Victim Loses


The attacker keeps the coins. The victim pays for power, hardware wear and cloud time, and gets slower systems. Because the goal is steady income, good cryptojacking is quiet. It tries to look like normal work, run at a low load and survive restarts. That is why malware of this kind can go unseen for weeks.


Cryptojacking, Mining and Trading Are Different


  • Legitimate mining: the owner chooses to run mining software on hardware they control and pay for.

  • Cryptocurrency trading: buying and selling coins. It uses no one else's computing power.

  • Cryptojacking: mining on someone else's resources without consent. The victim bears the cost.


A Simple Illustration


Picture a small design studio. One designer downloads a free codec pack from a search ad that looks real. A few days later the office desktops run hot, fans spin all day and renders take twice as long. No pop-up appears. A hidden program is using the graphics card in the background. This story is made up to explain the idea, but a documented 2026 campaign, covered later in this guide, followed a similar path.


How Does Cryptojacking Work?


Most cryptojacking follows the same five stages. The details change, but the pattern is steady.


Stage

What happens

What the victim may notice

1. Access

Malicious code gets a chance to run. This can come from a fake download, an unpatched service, stolen keys or a script on a web page.

Often nothing at all.

2. Payload

A miner is downloaded or started. Attackers often use known open-source miners such as XMRig, sometimes renamed.

A new or oddly named process.

3. Pool link

The miner connects to a mining pool. This is a server that hands out work and tracks each miner's share.

Steady outbound network traffic.

4. Persistence and evasion

The miner is set to return after a restart and to hide when someone looks. It may use scheduled tasks, startup entries, services or cron jobs.

The problem comes back after a reboot.

5. Payout

Earned coins go to a wallet the attacker controls.

Nothing. This stage is invisible to the victim.


CPU or GPU? Many miners use the processor. Others use the graphics card, because GPUs can do mining math faster. Microsoft's May 2026 research describes a campaign aimed at owners of powerful discrete GPUs. Its miners stopped when they detected user activity or heavy GPU use, which is exactly why a quiet check sometimes shows nothing. Source: Microsoft Security Blog.


Persistence and evasion. MITRE notes that some miners kill rival malware to keep the machine for themselves. Microsoft's 2026 report describes scheduled tasks, registry run keys and a startup shortcut that were recreated if removed, plus Microsoft Defender exclusions added for the miner files. The lesson is simple: ending one process may not end the problem.


What the attacker gains and what it costs you. The attacker's profit grows with every hour your system works for them. Your costs grow too: electricity, slower work, shorter hardware life and, in the cloud, a bill for compute you never wanted.


Types of Cryptojacking


The word covers several different situations. A browser tab that mines for ten minutes is very different from a hidden miner on a cloud cluster. This table shows the main differences.


Type

Where it runs

Common way in

Persistence

Best place to see it

Browser-based

Script inside a web page or extension

Compromised site, bad ad or extension

Usually only while the page or extension is active

Browser task manager, CPU use

Endpoint malware

Windows, macOS or Linux computers

Fake downloads, phishing, trojans

Scheduled tasks, startup items, services

Task Manager, Activity Monitor, endpoint security

Servers

On-premises or hosted servers

Unpatched or exposed services, weak passwords

Cron jobs, services, added SSH keys

top, logs, endpoint or SIEM alerts

Cloud VMs

Virtual machines in a cloud account

Stolen keys, exposed services

New instances or changed images

Billing, cloud audit logs, cloud threat detection

Containers and Kubernetes

Pods and containers

Exposed dashboards or APIs, bad images

New pods, images and jobs

Runtime tools, cluster logs, cloud alerts

GPU-based campaigns

PCs and servers with strong GPUs

Fake utilities, exposed AI or compute nodes

Same as endpoint or server cases

GPU usage graphs, endpoint tools


Browser Mining Is Not the Same as Persistent Malware


Browser mining was loud in 2017 and 2018. The best-known service, Coinhive, announced it would stop on March 8, 2019, saying the model was no longer economically viable. Source: The Hacker News. Browser mining still exists as a technique, but most modern campaigns that make news involve endpoints, servers and cloud accounts. Treat browser cases as real but usually shorter-lived. Details are in the browser section below.


What Changes by Platform


On a personal computer, the clues are heat, noise and slowness. On a Linux server, they are odd processes, cron entries and network connections. In the cloud, they are new resources, strange API activity and billing spikes. In containers, they are short-lived pods and images no one approved. Each place needs its own checks, which the later sections cover one by one.


How Do Cryptojacking Attacks Start?


A miner is only the last step. Before it runs, something has to give the attacker a way in. These are the common starting points. This section is for defense, so it names the paths without giving attacker steps.


  • Compromised websites or extensions. A hacked site or a rogue browser extension can load a mining script in your browser.

  • Deceptive downloads and poisoned search results. In a 2026 campaign, Microsoft linked more than 150 malicious domains to lookalike download pages for tools such as CrystalDiskInfo, HWMonitor and FurMark. Source: Microsoft Security Blog.

  • Phishing and [social engineering](@social-engineering). A link or attachment tricks a person into running a first-stage file.

  • Exploited vulnerabilities. Unpatched internet-facing software can let an attacker run code on a server.

  • Stolen or weak credentials. Leaked cloud keys and reused passwords let attackers log in instead of breaking in.

  • Cloud misconfigurations and exposed services. Open admin consoles and public APIs are easy to find with scanners.

  • Third-party software. Legitimate remote-access tools can be abused. Microsoft reported that the 2026 campaign silently installed ScreenConnect for persistent access.

  • Container images and dependencies. A public image or package can carry a miner that you pull in yourself.


Real Case: Tesla's Cloud, 2018


Researchers at RedLock reported that attackers found a Kubernetes console at Tesla that was not password protected. It exposed credentials for an AWS environment. The attackers mined cryptocurrency there and tried to hide by using an unlisted mining endpoint behind Cloudflare. The case shows how one open console can turn into a cloud-wide problem. Source: Engadget report on the RedLock findings.


Real Case: Malicious Docker Hub Images, 2021


Unit 42 at Palo Alto Networks found 30 malicious images from 10 Docker Hub accounts. Together they had over 20 million pulls. The researcher estimated about US$200,000 was mined, based on the half of the images that used a pool that shares this data. Source: Unit 42. The takeaway for Docker users is plain: pulling an image means trusting its author.


Hub-and-spoke illustration of an API connected to several services, each protected by a lock.

Your APIs and SaaS tools are part of your attack surface


Attackers look for unprotected endpoints, weak access checks and leaked keys. An authorized web app, API and SaaS security assessment looks for those problems from an outsider's point of view, so you can decide which exploitable weaknesses to fix first.




Cryptojacking Signs and Warning Symptoms


No single sign proves cryptojacking. Treat each one as a clue, then verify it. The table pairs every clue with a likely cause, a check and a harmless explanation.


Warning sign

What it could mean

How to verify

Common benign alternative

Sustained high CPU or GPU use while idle

A hidden miner is using the processor or graphics card

Find the process in Task Manager, Activity Monitor or top. Note when it started and where the file is

Updates, search indexing, a security scan, a game or render running in the background

Heat and constant fan noise

Steady heavy compute

Compare with the CPU and GPU graphs at idle

Dust, a hot room, charging or a long video call

Short battery life

Background load that never stops

Check the battery or Energy view for the top apps

Screen brightness, cloud sync or a new app

Slow apps and lag

Resources are being taken from your work

Compare with your normal baseline, including memory and disk

Low disk space, a memory leak or a pending update

Odd process trees or new startup tasks

Persistence set up by malware

Check the parent process, the file signature and scheduled tasks

IT management tools or software updaters

Unusual network traffic

Connection to a mining pool

Review destinations, ports and DNS lookups

Backups, video streaming or cloud sync

Security tool detections

A tool flagged a miner or related tool

Read the alert details, file path and quarantine action

A false positive on a legitimate tool you installed on purpose

Cloud resources you did not create

Stolen cloud credentials or an exposed service

Review audit logs for who created them, where and when

Autoscaling or a teammate's test

Billing anomalies

Stolen compute running in your account

Break down cost by service, region and day

A product launch, traffic spike or new project


Clues are not proof. Two or three clues that line up in time are much stronger than one. For example, a new scheduled task, a high-CPU process with a random name and traffic to an unknown server, all starting the same day, is a strong pattern. High CPU by itself is not.


What Are the Risks of Cryptojacking?


The direct harm is stolen computing power. Everything else depends on how the attacker got in and what else they did.


  • Electricity and compute costs. Mining runs the hardware hard for long periods. In the cloud, you pay for it by the hour.

  • Surprise cloud bills. One compromised account can start many large instances in a short time.

  • Performance and productivity. Slow computers and slow services cost staff time and can frustrate customers.

  • Service degradation. MITRE notes that resource hijacking can reduce the availability of systems or services.

  • Overheating and wear. Constant heavy load can raise temperatures and may shorten hardware life, though the size of the effect depends on cooling and hardware.

  • System compromise. A miner means someone already ran code on the system. Ask what else they could do with that access.


Mining versus other intruder actions. A miner by itself does not steal files or passwords. But the access that allowed it might allow other things. In the Tesla case, the same open console that allowed mining also exposed credentials to data storage. Treat data theft as a question to investigate, not as an assumption. Our guides on spyware, keyloggers and ransomware cover those other risks.


Documented Cases and Data


These are real, sourced examples. They are not a measure of how common cryptojacking is overall.


Case or data point

Year

What was reported

Source

Tesla Kubernetes console

2018

Unprotected console exposed AWS credentials. Mining ran in a Kubernetes pod

Coinhive shutdown

2019

In-browser Monero mining service ended on March 8, 2019

Docker Hub images

2021

30 malicious images, 10 accounts, over 20 million pulls, about US$200,000 estimated

TeamTNT

Tracked by MITRE

Deployed XMRig Docker images and infected containers and Kubernetes clusters

GPU mining campaign

2026

More than 150 linked domains since March 2026. ScreenConnect abused. GPU miners used


Real Case: The 2026 GPU Mining Campaign


Microsoft Defender researchers described victims who searched for utility software and landed on lookalike sites. A downloaded ZIP led to ScreenConnect, DLL sideloading and process hollowing. The malware then downloaded GPU miners such as gminer, lolMiner or SRBMiner-MULTI. It also created several persistence methods, added Defender exclusions and stopped mining when it saw user activity. Microsoft recommends cloud-delivered protection, EDR in block mode, network protection and web protection, and a specific attack surface reduction rule.


Cryptojacking vs. Legitimate Mining and Related Threats


The same program can be fine in one place and a threat in another. Consent, control and cost decide which it is.


Activity

Consent

Execution

Who controls it

Who pays

Right response

Legitimate mining

Yes. The owner chose it

Runs openly on owned hardware

The owner

The owner

Manage power, cooling and policy

Cryptojacking

No

Hidden and persistent

The attacker

The victim

Contain, investigate, remove and close the entry path

Conventional malware

No

Steals, locks or spies. Mining is not the main goal

The attacker

The victim

Incident response, based on what the malware does

Browser mining

Only if clearly disclosed and opt-in

Runs in a page or extension while it is active

The site or extension owner

The visitor

Close the page, remove the extension, block the script

Proxyjacking

No

Sells the victim's bandwidth instead of compute

The attacker

The victim

Same containment steps. Check network use


Do not judge by name or by pool connection alone. XMRig is open-source software that people use legally. A connection to a mining pool can also come from a hobby miner on your own network. What matters is whether you or your organization approved it.


Other threats can look similar. A rootkit may hide processes, a trojan may bring the miner in, and fileless malware may run it from memory. A mining payload can also be one part of a larger cyberattack.


Myths vs. Facts


Myth

Fact

Cryptojacking only happens in browsers.

Browser mining is one form. Modern reports also involve endpoints, servers, containers and cloud accounts.

It steals your passwords.

Mining itself uses compute. Credential theft needs separate malicious actions, though both can come from the same intrusion.

A miner always makes the computer run at 100%.

Many miners limit their load or pause when you are active to avoid notice.

Killing the process fixes it.

Persistence can bring it back, and the original entry point may still be open.

Only big companies are targeted.

Attackers use broad scans and fake downloads, so personal and small-business systems are also at risk.

A clean scan means a clean system.

A scan is one data point. Check persistence, accounts and network activity too.


How to Detect Cryptojacking: A Practical Workflow


Use this order when something looks wrong. It starts with read-only checks, so you do not change evidence by accident.


  1. Set a baseline. Know what normal looks like for this device or workload: usual CPU, GPU, memory, network and cloud spend.

  2. Match time to process. Note when the high load began. Find the process that uses the resources at that moment.

  3. Check origin, signature and parent. Where is the file? Is it signed by a known publisher? Which process started it?

  4. Look for persistence. Review scheduled tasks, startup items, services, cron jobs and new accounts or keys.

  5. Review security alerts. Check your antivirus, endpoint tool and cloud security findings for related events.

  6. Correlate network and cloud signals. Look for connections to unknown servers, strange DNS lookups, new cloud resources and cost jumps.

  7. Rule out benign workloads. Ask the owners. A build job, a render, a data task or an approved miner may explain it.

  8. Decide severity and escalate. Score the case and pass it on using your incident process.


What makes evidence strong? Look for several independent signs at once: an unapproved process, an unusual path or parent, new persistence, a pool or miner-like connection, and a start time that matches another event. If you are unsure, do not delete anything. Isolate the device from the network if you can do so safely, capture notes and screenshots, and ask your security team or a qualified responder.


How to Check for Cryptojacking on Windows


These steps work on most Windows PCs. On a work device, follow your company's policy first.


  • Task Manager. Press Ctrl+Shift+Esc. Open the Processes tab and sort by CPU, then by GPU. Right-click a column header to add GPU and Power usage columns if needed. Use the Performance tab to see overall load.

  • Microsoft Defender. Open Windows Security, then Virus & threat protection. Run a scan and review Protection history. For a deeper check, use the Microsoft Defender Offline scan option.

  • Process Explorer. Microsoft Sysinternals offers Process Explorer. It shows open handles and loaded DLLs for a selected process, and lists processes with their owning accounts. This helps you judge whether a process is what it claims to be.

  • Startup and tasks. In Task Manager, open the Startup apps tab. Open Task Scheduler and review tasks with vague names, such as health or cache names, that point to unusual folders.

  • Exclusions. In Windows Security, open Manage settings and review Exclusions. Microsoft reported that the 2026 campaign added Defender exclusions for its own files, so unknown exclusions deserve attention.

  • Origin and signature. Right-click the file, open Properties and check the Digital Signatures tab and the folder path. A system-sounding name in a user folder is a red flag.


Do not blindly end unknown processes or delete files. Some are needed by Windows. Others may be evidence. Note the name, path and time, then follow the steps in the removal section.


How to Check on macOS


macOS gives you simple tools. They are enough for a first look.


  • Activity Monitor. Open it from Utilities. In the CPU tab, sort by % CPU. In the Energy tab, look for apps with high energy impact and for apps that keep the Mac awake. Check the GPU view if your Mac shows it.

  • Login and background items. Open System Settings, then General, then Login Items & Extensions. Remove nothing yet. Write down items you do not recognize.

  • Browser extensions. Review the extensions in each browser. Remove ones you did not choose or no longer need, after you have noted their names.

  • Built-in protection. macOS has built-in protections such as Gatekeeper and XProtect. Keep the system updated so these stay current.

  • Managed Macs. On a work Mac, an admin may control profiles and security tools. Report findings instead of changing settings yourself.


How to Check on Linux and Servers


On Linux servers, use read-only commands first. Only run them on systems you are authorized to inspect. They look but do not change anything.


  • top or htop shows live CPU and memory use. In top, press P to sort by CPU.

  • ps -eo pid,ppid,user,%cpu,etime,cmd --sort=-%cpu | head -15 lists the busiest processes with their parent ID, owner and run time.

  • ls -l /proc/PID/exe shows the real file behind a process ID. Replace PID with the number you saw.

  • systemctl list-timers --all and systemctl list-units --type=service --state=running show scheduled timers and running services.

  • crontab -l shows the current user's cron jobs. Also review the system cron folders and each user's authorized_keys file.

  • ss -tunp lists network connections with their processes. Look for unfamiliar remote servers, especially from busy processes.

  • journalctl and the files in /var/log help you see logins, service changes and sudo use around the start time.


Legitimate versus unauthorized work. Backups, builds, video encoding and data jobs can use lots of CPU. Compare the process owner, schedule and change tickets. If nobody can explain the job, the path is unusual or the process talks to an unknown server, escalate. Open-source monitoring such as Wazuh can watch for changes like new SSH keys and cron edits. Also see our guide to the Linux kernel for background.


Detecting Browser-Based Cryptojacking


Browser cases are usually easier to test, because closing the page often stops the load.


  • Use the browser's task manager. In Chrome and Edge, press Shift+Esc on Windows and Linux, or use the browser menu's task manager entry. It lists each tab and extension with its CPU use. Firefox offers about:processes.

  • Find the tab or extension. A single tab or extension that stays high while the page is idle deserves a closer look.

  • Test by closing it. If CPU use drops right after you close one page or disable one extension, you have a good lead.

  • Audit extensions and permissions. Remove extensions you do not use. Review site permissions and notification access.

  • Look at the page itself. Use your security tool or a browser feature that blocks known mining scripts. Note that script blocking tools may not catch everything.


Do not equate a heavy page with mining. Games, video calls, maps and web apps can use lots of CPU too. You need a pattern: high load tied to one page or extension, no clear reason, and relief when you close it.


Detecting Cryptojacking in AWS, Google Cloud, and Kubernetes


In the cloud, the first alert is often a bill or a notification, not a slow laptop. Watch for these signs:


  • A sudden rise in instances or containers, especially in regions you do not use.

  • CPU, GPU or network spikes on workloads that were quiet before.

  • New IAM users, roles or access keys, or API calls from unfamiliar places.

  • Runtime processes that match miner behavior, and outbound connections to mining pool ports or domains.

  • Cost and usage jumps in a service you rarely use.


Amazon Web Services and GuardDuty


In AWS, GuardDuty has finding types that start with CryptoCurrency:. These include EC2/BitcoinTool.B, EC2/BitcoinTool.B!DNS, Lambda/BitcoinTool.B and Runtime/BitcoinTool.B. They depend on data sources: the EC2 findings use VPC flow logs or DNS logs, and the Runtime findings need GuardDuty Runtime Monitoring on the workload. Source: Amazon GuardDuty finding types. AWS's own guidance, updated in 2026, also recommends enabling GuardDuty across accounts and Regions, using budgets and CloudWatch alarms, restricting outbound traffic and using least-privilege IAM. Source: AWS Security Blog.


Google Cloud


Google's Security Command Center best-practices page names Premium or Enterprise tier use, with Event Threat Detection and VM Threat Detection, as the base for cryptomining detection. VM Threat Detection works from the hypervisor, needs no agent, and scans memory after VM creation and then every 30 minutes. Its cryptomining detectors use hash and YARA matches. Findings include Execution: Cryptomining YARA Rule and Execution: Cryptomining Hash Match. Limits apply: Windows coverage for miners is limited, and some disk and VM types are not scanned. Sources: best practices and VM Threat Detection overview.


Kubernetes and Containers


In Kubernetes and other container platforms, use runtime detection, audit logs and network policies. Falco, an open-source runtime security tool, documents rules for miner pool ports and the stratum protocol. Source: Falco documentation blog. Elastic also publishes prebuilt rules for cryptomining, including a rule for suspicious mining service creation on Linux. Source: Elastic detection rules.


Limits to remember. No tool sees every miner on every platform. Findings can depend on logs, feature settings, subscription tier and region. Use detection together with budgets, logging and good cloud governance. If you suspect abuse, preserve logs and snapshots before you delete instances, and follow your incident process.


Best Cryptojacking Detection Tools Compared


There is no single best tool. Process viewers show load. Antivirus and endpoint tools block and alert. SIEM and runtime tools correlate events. Cloud services watch accounts and workloads. This table compares what each is for. Features vary by subscription, configuration, region, operating system and updates, and we did not run first-hand tests.


Tool

Platform

What it detects or helps investigate

Best for

Important limitation

Cost model

Windows Task Manager

Windows

Live CPU, GPU and memory use by process

Individuals doing a first check

Shows load only, not intent or origin

Included with Windows

Microsoft Defender

Windows. Defender for Endpoint covers more platforms

Known miners, for example Trojan:MSIL/CoinMiner!MS, plus behavior alerts in the endpoint product

Individuals (built-in antivirus) and IT teams (endpoint product)

Attackers may try to add exclusions. Coverage varies by setup

Antivirus is built in. Endpoint product is licensed

Sysinternals Process Explorer

Windows

Handles, loaded DLLs and process owners

Analysts and IT staff

A process viewer, not antivirus. Needs skill to read

Downloaded from Microsoft. No price is stated on the page

macOS Activity Monitor

macOS

CPU, Energy and per-process use

Mac users

Shows load only

Included with macOS

Linux tools: top, htop, ps, ss, journalctl

Linux

Live processes, connections and logs

Administrators

Manual. No alerts by itself

Included or open-source packages

Wazuh

Linux, Windows and macOS agents

File changes, log analysis and custom rules, such as new SSH keys and cron edits

Teams wanting open-source monitoring

Needs setup, tuning and custom rules

Open-source platform. Support options vary

Elastic Security

Multi-platform

Prebuilt rules, including suspicious mining service creation on Linux and an AWS ECS cryptomining rule

Security operations teams

Needs the right data sources, such as Elastic Defend or CloudTrail

Tiers and subscriptions vary

Falco

Linux hosts, containers and Kubernetes

Runtime rules for miner ports and the stratum protocol

Cloud-native teams

Rules need tuning. Some are off by default

Open-source runtime security project

AWS GuardDuty

AWS

CryptoCurrency findings for EC2, Lambda and runtime activity

AWS cloud teams

Needs the right data sources. May miss miners using unlisted pools

Usage-based AWS service

Google Cloud Security Command Center

Google Cloud

Memory-based miner detection through VM Threat Detection and event findings

Google Cloud teams

Limited Windows miner coverage. Some VMs and disks not scanned

Premium tier required

Enterprise EDR (category)

Endpoints and servers

Process behavior, network activity, hunting and response actions

Organizations with security staff

Cost, tuning and coverage gaps. Needs people to act on alerts

Commercial subscription


Facts in this table come from vendor and project documentation: Microsoft, Process Explorer, Wazuh, Elastic, Falco, AWS and Google Cloud.


Which Option Fits Whom


  • Individual: built-in antivirus, Task Manager or Activity Monitor, and careful downloads.

  • IT team: endpoint protection, central logging and a SIEM or log platform.

  • Cloud team: native services such as GuardDuty or Security Command Center, plus budgets, audit logs and runtime tools for containers.

  • Enterprise: an EDR platform and a managed or in-house threat detection and response process.


Why penetration testing is not a detection tool. A penetration test is a point-in-time, authorized check of your own systems for exploitable weaknesses. It does not watch your environment day to day and does not directly detect a miner. Use detection tools to see activity. Use testing to find the weaknesses that might let it in.


Illustration of security findings ranked by severity bars, from low to critical.

Monitoring finds activity. Testing finds weaknesses.


Detection tools tell you when something suspicious is happening. Independent, authorized application testing answers a different question: which security weaknesses should you fix before they are used? Many teams need both views.




How to Remove Cryptojacking Safely


The right steps depend on where the miner runs. A common rule applies everywhere: contain first, keep evidence, then clean up and close the entry point. Avoid destructive one-line fixes, deleting logs or switching off security tools.


  • Personal device: disconnect from the network, update the system and security software, run a full scan with a trusted tool, and follow its quarantine or removal advice. Review startup items, scheduled tasks and extensions. Change passwords from a clean device and turn on multi-factor authentication. If the miner returns, consider a clean reinstall.

  • Browser only: close the tab, remove the suspicious extension, update the browser and clear site data for the offending site.

  • Managed work endpoint: report it to IT or security. Let them isolate the device with their tools and collect evidence.

  • Linux server: snapshot the disk and save logs first. Isolate the host, review accounts, SSH keys, cron jobs and services, then patch the entry path. Rebuild from a trusted image when you cannot be sure it is clean, and rotate secrets.

  • Cloud workload: preserve audit logs and snapshots, disable exposed keys and roles, stop or isolate affected instances in every region, then replace workloads from trusted images. Add budget alerts before you resume.


Why killing a process is not enough. Persistence can restart the miner, and the original weakness may still be open. Verify the fix: monitor CPU, GPU, network and cost for a while, repeat scans and check that tasks, keys and accounts you removed do not return. One clean scan does not prove a system is clean. For business incidents, use your approved incident response process.


Preventing Cryptojacking on Personal Devices


Start with the controls that block the most common paths. Each one targets a specific problem.


  • Install updates quickly. Patches close holes that attackers scan for.

  • Download from official sources. Type the vendor's address yourself instead of clicking a search ad. This stops lookalike download pages.

  • Keep few browser extensions. Fewer extensions means fewer places for hidden scripts.

  • Use trusted security software. Keep real-time protection and cloud protection turned on. See our guide on adware and the computer virus overview for related threats.

  • Use strong sign-in. A password manager and multi-factor authentication protect your accounts if a device is hit.

  • Use a standard account day to day. Less privilege makes persistence harder.

  • Keep backups. If you must reinstall, you will not lose your files.

  • Learn the warning signs. Fan noise, heat and slowness are worth a quick check.


Preventing Cryptojacking in Businesses, SaaS, and Cloud


Layer your controls by phase: stop the way in, limit what a stolen account can do, watch compute and cost, detect strange execution, contain it fast, then recover and retest.


Priority

Control

What it addresses

Phase

1

Patch management and asset inventory

Known vulnerabilities and unknown systems

Prevent initial access

2

MFA and least-privilege IAM

Stolen or weak credentials

Reduce privilege

3

Secrets protection and key rotation

Leaked cloud keys and tokens

Reduce privilege

4

Web, API and SaaS security reviews, including penetration testing and vulnerability assessment

Exploitable application weaknesses

Prevent initial access and verify

5

Trusted container images and registry rules

Poisoned or unknown images

Prevent initial access

6

Egress controls where practical

Miner connections to pools and control servers

Detect and contain

7

Logging, budgets and cost alerts

Unexpected compute and spend

Monitor

8

EDR, runtime and cloud threat detection

Suspicious execution

Detect

9

Incident playbooks and threat hunting

Slow, unclear response

Contain and recover

10

Retesting after fixes

Fixes that did not hold

Recover and retest


Continuous monitoring versus point-in-time assessment. Monitoring runs all the time and looks for activity. A penetration test or vulnerability assessment looks at a moment in time for weaknesses. You need monitoring to catch what happens, and assessments to find and fix what makes it possible. Useful background: network security, cloud-native security, API basics, firewalls, cloud compliance and cloud spending statistics.


When to Seek Professional Security Help


Your situation

Who can help

Why

Active attack or signs of ongoing intrusion

Incident response and forensics provider

They contain the attack and preserve evidence

Suspected infection on an endpoint

Your EDR provider or a qualified cleanup team

They can confirm the infection and remove it correctly

Cloud abuse or surprise cloud bills

Cloud security or cloud incident response specialists

They know the provider's logs and controls

Weak application, API or SaaS security

An authorized, scoped security assessment

It finds weaknesses before attackers do

Need to confirm fixes worked

Retesting, separately scoped

It checks that issues were actually closed


If your concern is application, API or SaaS exposure, Articsledge offers scoped, authorized web app, API and SaaS security testing and vulnerability assessment, with findings and reporting. Separately scoped retesting may also be available. Methods, scope, deliverables and terms are agreed first. This is a security assessment service. It is not incident response, malware removal or round-the-clock monitoring, and it cannot promise to prevent every attack. Learn more in our penetration testing guide or on the penetration testing services page.


Illustration of a security report page with a circular retest arrow and a shield, representing findings and remediation verification.

Fixes you can prove


Evidence-backed testing gives your team clear findings it can act on. After fixes, a separately scoped retest can check that the issues were really closed. Scope, methods and terms are agreed before work starts.




FAQ


What is cryptojacking in simple words?


It is when someone secretly uses your computer, server or cloud account to mine cryptocurrency. They keep the coins. You pay with slower systems, higher power use and sometimes bigger cloud bills.


Is cryptojacking malware?


Usually yes, when it runs without permission. The miner itself may be a normal open-source program, but it becomes malicious when it is installed or run without authorization. Learn more in our malware guide.


How is cryptojacking different from legitimate mining?


Consent. A legitimate miner is chosen and paid for by the owner of the hardware. Cryptojacking runs on someone else's resources without approval, and the victim pays the costs.


What are the signs of cryptojacking on a CPU?


Common clues include steady high CPU use when idle, loud fans, heat, short battery life and slow apps. These signs have many harmless causes, so confirm the process, its origin and its network activity.


Does high CPU usage always mean cryptojacking?


No. Updates, scans, indexing, games, builds and renders also use lots of CPU. You need a pattern, such as an unknown process, new persistence and mining-pool-like traffic.


Can a website mine cryptocurrency in my browser?


Yes, in theory and in past campaigns. A page script can use your CPU while the page is open. Coinhive, a well-known service, shut down in 2019. Use your browser's task manager to find a tab with unexplained load.


How do I check for cryptojacking on Windows?


Open Task Manager and sort by CPU and GPU. Check the file path and signature of any odd process, review startup apps and scheduled tasks, and run a Microsoft Defender scan. Do not delete files you do not understand.


Can Macs and phones be cryptojacked?


Yes. MITRE lists macOS examples such as CookieMiner. Phones can be affected by malicious apps, though in 2018 Apple and Google restricted mining apps and extensions in their stores. That lowers the risk but does not remove it.


Which tool is best for detecting cryptojacking?


It depends on where you work. Task Manager helps individuals, endpoint tools help IT teams, and services such as GuardDuty or Security Command Center help cloud teams. Many organizations need more than one layer.


How does cryptojacking affect cloud bills?


Attackers who get cloud access can start many large instances or containers. You pay for that compute by the hour. Budgets, cost alerts and audit logs help you spot it early.


Can cryptojacking damage hardware?


Constant heavy load raises heat and may shorten the life of poorly cooled hardware. The real effect depends on cooling and the device. Modern systems usually slow down or shut off to protect themselves.


Does cryptojacking steal data?


Not by itself. Mining uses compute power. But the access used to install a miner may allow other actions, so you should check for credential theft and other changes during an investigation.


Is cryptojacking illegal?


Using other people's computers without permission is illegal in many places, but laws differ by country and region. This is general information, not legal advice. Ask a qualified lawyer about your situation.


How do I remove cryptojacking?


Contain the device, keep evidence, run trusted scans, remove persistence and extensions, change passwords and fix the entry point. For business systems, use your incident response process. See the removal section.


Can penetration testing help prevent cryptojacking?


It can help indirectly. An authorized penetration test finds exploitable weaknesses in your web apps, APIs and SaaS tools that attackers might use. It does not detect miners or replace monitoring.


Key Takeaways


  • Cryptojacking is mining without permission. Authorization is the dividing line.

  • The direct harm is stolen compute, which shows up as slowness, power use and cloud bills.

  • High CPU is a clue. Strong evidence needs the process, its origin, persistence and network activity.

  • Browser mining exists but is mostly a historical pattern. Modern cases often involve endpoints, servers and cloud accounts.

  • Cloud and container cases need logs, budgets, IAM controls and runtime detection.

  • Process viewers, antivirus, EDR, SIEM, runtime and cloud tools do different jobs.

  • Contain and preserve evidence before you clean. Killing one process may not end the problem.

  • Prevent by closing entry points: patching, MFA, careful downloads, least privilege and tested applications.


Actionable Next Steps


If you suspect an active attack on a business system, stop and escalate to your incident response team or provider first. Do not wipe logs or rebuild systems before evidence is saved.


For Individuals


  1. Open Task Manager or Activity Monitor and check what uses CPU and GPU when you are idle.

  2. Update your operating system, browser and security software.

  3. Review startup items, scheduled tasks and browser extensions. Write down anything unfamiliar.

  4. Run a full scan with trusted security software and follow its guidance.

  5. Turn on multi-factor authentication and change important passwords from a clean device.


For Business and Cloud Administrators


  1. Set budgets and cost alerts, and turn on audit logging in every account and region.

  2. Enable cloud threat detection and runtime monitoring where your plan supports it.

  3. Review IAM roles, remove unused keys and require MFA.

  4. Build an inventory of internet-facing systems, and patch them first.

  5. Test your web apps, APIs and SaaS tools with an authorized, scoped assessment, then retest after fixes.


Glossary


  • Cryptojacking: secret use of someone else's computing resources to mine cryptocurrency.

  • Compute hijacking: MITRE's name for using compromised systems' computing power for tasks the attacker wants.

  • Mining pool: a server that shares mining work among many miners and tracks their shares.

  • XMRig: a popular open-source miner that is used both legitimately and by attackers.

  • GPU: graphics processing unit. A chip that can do mining math quickly.

  • Persistence: a method that keeps malware running after a restart or removal attempt.

  • EDR: endpoint detection and response, a security tool that watches device behavior and helps respond.

  • IAM: identity and access management, the rules about who can do what in a system or cloud account.

  • Indicator of compromise: a clue, such as a file hash or domain, that suggests a system was attacked.

  • Container: a packaged, isolated environment for running software.

  • False positive: an alert or suspicion that turns out to be harmless.


Sources & References



bottom of page