control-flow hijack

Concept

A control-flow hijack is a security violation in which an attacker causes a program or processor to execute at a control-flow location not intended by its software or instruction set, diverting execution toward attacker-chosen targets. The concept applies across software contexts (LLM-based multi-agent systems), binary exploitation (memory-corruption attacks and speculative execution), and hardware (microarchitectural bugs in CPU designs that allow operand data to steer the architectural program counter).

First seen 7/6/2026

Last seen 7/6/2026

Evidence 4 chunks

Wiki v1

WIKI

control-flow hijack

Definition

A control-flow hijack is a security violation in which the control flow of a program, agent, or processor is redirected to a target chosen by an attacker instead of one intended by the legitimate software, instruction set, or orchestration policy. The hijack causes execution to deviate from architecturally valid or policy-valid control flow, and it is one of the canonical primitives of binary and systems security alongside information leakage.

READ FULL ARTICLE →

NEIGHBORHOOD

No graph connections found for this entity yet. It may appear in future ingestion runs.

explore full graph →

CITATIONS

6 sources

6 citations — click to expand

[1] A control-flow hijack manipulates orchestration in multi-agent systems so agents perform unsafe actions, and defenses such as LlamaFirewall use alignment checks of inter-agent communications. Breaking and Fixing Defenses Against Control-Flow Hijacking in Multi-Agent Systems

[2] Memory-safety defenses (stack canaries, CFI, Go bounds checks) can be bypassed using speculative control-flow hijacks, including speculative ROP via stitched gadgets, with software mitigations available at moderate performance cost. Bypassing memory safety mechanisms through speculative control flow hijacks

[3] Microarchitectural CFI (μCFI) enforces that the microarchitectural control flow is only influenced by instructions for which the ISA explicitly allows a control or data path from their operands to the PC, and its violation by attacker-controlled data yields a control-flow hijack. μCFI: Formal Verification of Microarchitectural Control-flow Integrity

[4] Information flows verified by μCFI cover four cases, including invalid architectural CF driven by attacker-controlled data, which corresponds to a control-flow hijack. μCFI: Formal Verification of Microarchitectural Control-flow Integrity

[5] Kronos CVE-2024-44927 is a control-flow hijack where, with a memory that does not always acknowledge requests in the next cycle, a jalr or branch reads from the previous instruction's input register and can be steered by attacker-controlled data. μCFI: Formal Verification of Microarchitectural Control-flow Integrity

[6] Kronos CVE-2023-51973 is a control-flow hijack in which an addi, slti, or sltiu data operand can be copied into the mtvec CSR, allowing a user-mode attacker to influence the architectural CF in machine mode. μCFI: Formal Verification of Microarchitectural Control-flow Integrity