macOS Dylib Hijacking Detection via Endpoint Security
A DEF CON 34 talk details how macOS dylib hijacking and dylib-based malware evade process-level security tools, and how Endpoint Security mmap events can detect and block them.
Sourceobjective-see.org — open the original
What to check
- Applies to
- Whether this concerns you is not assessed; the names the report mentions are listed below.
This Objective-See blog post recaps a DEF CON 34 talk examining dylib-based attacks on macOS, arguing that despite mitigations up through macOS 26, dylib hijacking remains possible in limited forms. The author reviews historical and real-world examples of dylib abuse: Flashback malware injecting via DYLD_INSERT_LIBRARIES into Safari for ad fraud, evidence suggesting Equation Group's OS X implant supported a dylib-loaded variant, the 3CX supply chain attack which trojanized libffmpeg.dylib to download second-stage payloads, and a SIP-bypass exploit (documented by researcher CodeColorist) that abused a vulnerable dylib lookup path (DEVELOPER_DIR) in an Apple-signed sampling tool holding the com.apple.system-task-ports entitlement. The core thesis is that macOS security decisions (entitlements, TCC, Endpoint Security attribution, firewall rules) are made at the process level, so malicious code loaded as a dylib inherits a host process's trust while remaining invisible to tools like LuLu or eslogger, which attribute all activity to the host application (e.g., Photoshop) rather than the injected library.
The post then details defensive techniques: static enumeration of Mach-O LC_LOAD_DYLIB dependencies (as used in the open-source Dylib Hijack Scanner) to spot duplicate @rpath dependencies indicating a hijack; runtime enumeration of loaded libraries in a remote process via proc_pidinfo with PROC_PIDREGIONPATHINFO, which avoids the now-restricted task_for_pid approach; and load-time monitoring/blocking using Apple's Endpoint Security ES_EVENT_TYPE_AUTH_MMAP event, which allows a security tool to inspect and deny malicious executable mappings before they load, effectively preventing dylib hijacking and other dylib-based payloads. Limitations are noted, including incomplete visibility into the dyld shared cache, potential TOCTOU issues between mapped memory and on-disk files, and the need to also watch ES_EVENT_TYPE_AUTH_MPROTECT for more exotic in-memory techniques. The article is defensive research and detection guidance, not a report of a new active campaign.
Mentioned in this report
Source reporting: https://objective-see.org/blog/blog_0x89.html
Free public brief. Slack alerts on the vendors you run come with the paid plans, and a new account gets three days of them, no card.
What this brief leaves out
This page is the free tier: the write-up, the severity, the names. In the app the same report carries live profiles of the actors, malware, and campaigns it names. The paid plans add the security research and news reporting the free tier leaves out.
It sits in a corpus of 16,433 reports from 149 sources, 661 of them written in the last seven days, and it grows through the day.
A new account starts with three days of all of it, no card, then keeps the government and CERT reporting free. Slack alerts are a paid-plan feature: name the vendors you run and Slack hears about the reports that touch them, not the rest of the week's. Nothing is charged unless you subscribe.
Create a free account What it costs