Data Exfiltration Detection


Data exfiltration is the unauthorized transfer of data from an organization to an external destination controlled by an adversary.

  • can be from insider or malware

Data Exfiltration Phases

  • Discovery / Collection
    • attacker locates sensitive files 
  • Staging / Compression
    • attacker aggregates, compresses, encrypts, or encodes files (ZIP, RAR, 7z, tar, base64, steganography)
  • Exfiltration transport
    • transfer over network, removable media, cloud, or covert channels
  • Command & Control (C2) coordination
    • orchestrates transfer and confirms receipt

Techniques and Indicators

  • Requires correlating host and network-level indicators such as:
    • unusually large or frequent outbound uploads (proxy/firewall)
    • long or high-entropy DNS queries
    • suspicious process command-lines and network connections (Sysmon/EDR)
    • cloud storage API activity
    • removable-media events
  • SOC L1 triage focuses on:
    • source host/user
    • destination
    • transferred volume
    • process identity/command-line
    • and supporting evidence across proxy, DNS, flow, host, and cloud logs
TechniquesExamplesIndicator of Attack & where to look
Network-basedHTTP/HTTPS uploads to S3/Azure Blob/webmail, FTP/SFTP/SCP, DNS tunnelling, ICMP/covert protocols, custom TCP/UDP.Proxy/web gateway logs (large POSTs, uploads to cloud endpoints), firewall/NGFW flows (high bytes to single IP/ASN), netflow (spikes/outbound flows), DNS logs (long hostnames, TXT queries).
Host-basedPowershell/Invoke-WebRequest, rclone, awscli, curl/wget, archive creation (zip/rar), use of removable USBs, ADS/hidden streams.Sysmon/EDR (Process Create, Network Connect, File Create events), Windows Security (4663/4656 object access), auditd/shell history on Linux, and removable-media events.
Cloud exfiltrationS3 PutObject / multipart upload, Azure Blob uploads, Google Cloud Storage objects. Insert, Drive/SharePoint external sharing.CloudTrail / Azure Activity / GCP Audit, cloud storage access logs, unusual service-account or IP activity.
Covert & encodingDNS tunnelling, base64 or chunked encoding, steganography into images/audio, splitting files into many small requests (low-and-slow).DNS logs, proxy logs with many small POSTs, correlation of intermittent uploads + suspicious process activity.
Insider & collaboration toolsSlack/Teams/Dropbox/Google Drive/Box uploads or sharing to external users; compromised employee accounts.Audit logs (share events, file downloads), and mail logs.
General IoAs & triage signalsA large outbound volume to external IPs/domains, unknown destination domains, suspicious processes/command lines, many file read events followed by an outbound connection, and multipart/streamed uploads.Correlate: Proxy/Firewall/Netflow, DNS, Sysmon/EDR (EventID 1/3/11), mail server logs.

Methods

Reference

Shabbir, S. (n.d.). Data Exfiltration Detection. TryHackMe. Retrieved June 9, 2026, from https://tryhackme.com/room/dataexfildetection