Skip to main content
Process diagram: the data breach path behind this report, with the fault at Unsupervised ML. Illustration: SecurityXP

FlowSH Correlates Anomalous NetFlows With Server-Side Logs to Catch SQLi That ModSecurity Misses

· 1 min read · SecurityXP

Key Facts

  • To access encrypted payloads, some NIDS resort to TLS-interception proxies, which leads to data leaks and security risks.
  • The system combines unsupervised machine learning to flag anomalous NetFlows and similarity search over server-side logs using Locality Sensitive Hashing (LSH) to confirm attacks.
  • We implemented and evaluated FlowSH in a testbed with multiple web scanners and demonstrated its detection effectiveness, detecting 70% of SQLi attacks missed by ModSecurity, a leading web application firewall.
  • Among web attacks, SQL injection (SQLi) remains a persistent, damaging web applications connected to databases.
  • Computer Science > Cryptography and Security [Submitted on 18 Nov 2026] Title:FlowSH: Detecting Web Attacks by Correlating Anomalous Network Flows and Endpoint Logs View PDF HTML (experimental)Abstract:Web applications compose the core of internet services, rendering them prominent targets for audacious attackers.

FlowSH, a detection system posted 18 November 2026 under Computer Science > Cryptography and Security, combines unsupervised machine learning on network flows with similarity search across server-side logs. In a testbed evaluation using multiple web scanners, it detected 70% of SQL injection attacks that ModSecurity, described in the paper as a leading web application firewall, missed.

Two Telemetries, One Verdict

FlowSH’s first stage uses unsupervised machine learning to flag anomalous NetFlows. Its second stage confirms candidate attacks through Locality Sensitive Hashing (LSH) similarity search over endpoint logs. The design sidesteps a shortcut some network intrusion detection systems take: TLS-interception proxies that expose encrypted payloads, which the paper states leads to data leaks and security risks. FlowSH reaches its verdict on flow anomalies plus server-side evidence instead of decrypted traffic.

The Attack Class

The paper positions web applications as the core of internet services and prominent targets for attackers. SQL injection remains a persistent and damaging attack class against web applications connected to databases, the class FlowSH was built and tested against. The 70% detection figure comes from SQLi attempts ModSecurity missed across a testbed running multiple web scanners, where flow-level anomalies flagged by the unsupervised model were confirmed against endpoint logs by LSH similarity search.

Tags: Application Security NIDS NetFlows Locality Sensitive Hashing FlowSH SQLi ModSecurity

Sources & References

S SecurityXP
SecurityXP Cybersecurity News & Analysis

SecurityXP delivers cybersecurity news, vulnerability analysis, data breach reports, and threat intelligence.

Security Digest

The most important cybersecurity stories, CVE alerts, and threat intelligence — one concise email. Unsubscribe anytime.

Related Articles

Process diagram of the data breach path: Master key made through AWS role assume, with the fault at Master key made.
AI/ML Security

How attackers are jailbreaking LLMs with CTF framing and how to catch them

Over the past 30 days, we’ve collected data from other source IPs that validate our jailbreaking theory: 159.89.93.86 created a LiteLLM master-scoped API key...

Process diagram of the exploitation chain: Agent session through Missed transfer, with the fault at Missed transfer.
AI/ML Security

Auditing the State Transfers That Execution Traces Miss in Long-Horizon LLM Agents

A paper submitted on 17 Nov 2026 to Computer Science Cryptography and Security, “Securing and Auditing Long-Horizon LLM Agents: State-Transfer Auditing…

Process diagram of the data breach path: Agent spawned through Data exposed, with the fault at Shadow access.
AI/ML Security

Every AI Agent Is an Identity. Most Organizations Don't Treat Them That Way

This is no longer theoretical, 65% of organizations experienced a security incident involving an AI agent in the past year, with 61% reporting exposure or...

Process diagram of the data breach path: OVHai preview through Beta at Summit, with the fault at Partner apps.
AI/ML Security

OVHcloud previews AI workspace with encrypted tools

OVHcloud says OVHai Workspace includes an end-to-end encryption option covering data and communications, including within partner applications integrated...