Trojan Source: Invisible Vulnerabilities
Not on the map. Why?
Abstract (arXiv)
We present a new type of attack in which source code is maliciously encoded so that it appears different to a compiler and to the human eye. This attack exploits subtleties in text-encoding standards such as Unicode to produce source code whose tokens are logically encoded in a different order from the one in which they are displayed, leading to vulnerabilities that cannot be perceived directly by human code reviewers. 'Trojan Source' attacks, as we call them, pose an immediate threat both to first-party software and of supply-chain compromise across the industry. We present working examples of Trojan Source attacks in C, C++, C#, JavaScript, Java, Rust, Go, Python, SQL, Bash, Assembly, and Solidity. We propose definitive compiler-level defenses, and describe other mitigating controls that can be deployed in editors, repositories, and build pipelines while compilers are upgraded to block this attack. We document an industry-wide coordinated disclosure for these vulnerabilities; as they affect most compilers, editors, and repositories, the exercise teaches how different firms, open-source communities, and other stakeholders respond to vulnerability disclosure.
Key facts. Cited by 8 papers in the map (data through 7 Oct 2026). First posted on arXiv 30 Oct 2021.
Builds on
No reference list yet.
Built on by
Citing papers per year: 2024: 2; 2025: 1; 2026: 5.
8 papers.
- 2024Architectural Neural Backdoors from First Principlescited by 10
- 2024Impeding LLM-assisted Cheating in Introductory Programming Assignments via Adversarial Perturbationcited by 5
- 2025RAG-Pull: Turning Retrieval into a Code-Injection Channel via Invisible Unicode Perturbationscited by 4
- 2026Hidden-in-Plain-Text: A Benchmark for Social-Web Indirect Prompt Injection in RAGcited by 3
- 2026An Application-Layer Multi-Modal Covert-Channel Reference Monitor for LLM Agent Egresscited by 2
- 2026Reverse CAPTCHA: Evaluating LLM Susceptibility to Invisible Unicode Instruction Injectioncited by 0
- 2026Multi-Layer Context Camouflaging: A Semantic Superposition and Contextual Lamination Framework for Malpractice-Resilient Online Assessmentcited by 0
- 2026Subjects, Not Authors: The Authorship Hazard in Agentic Dataspacescited by 0
- 2026Subjects, Not Authors: The Authorship Hazard in Agentic Dataspaces24 Sep 2026
- 2026Multi-Layer Context Camouflaging: A Semantic Superposition and Contextual Lamination Framework for Malpractice-Resilient Online Assessment13 Aug 2026
- 2026An Application-Layer Multi-Modal Covert-Channel Reference Monitor for LLM Agent Egress20 May 2026
- 2026Reverse CAPTCHA: Evaluating LLM Susceptibility to Invisible Unicode Instruction Injection26 Feb 2026
- 2026Hidden-in-Plain-Text: A Benchmark for Social-Web Indirect Prompt Injection in RAG16 Jan 2026
- 2025RAG-Pull: Turning Retrieval into a Code-Injection Channel via Invisible Unicode Perturbations13 Oct 2025
- 2024Impeding LLM-assisted Cheating in Introductory Programming Assignments via Adversarial Perturbation12 Oct 2024
- 2024Architectural Neural Backdoors from First Principles10 Feb 2024