Impact

Data security Vulnerabilities

Issues affecting data confidentiality and integrity

5 matching entries out of 153 in this category

Data security research results

5 entries

CoT Detector Obfuscation Bypass

Updated 12/30/2025

LLM-based code agents and vulnerability detectors employing Chain-of-Thought (CoT) reasoning are susceptible to automated adversarial code obfuscation. The vulnerability exists because CoT mechanisms expose the model's decision logic, allowing reinforcement learning frameworks (such as CoTDeceptor) to iteratively refine code transformations based on the detector's own reasoning traces. By optimizing for "reasoning instability" and "hallucination" rather than just syntactic evasion, attackers…

CoTDeceptor: Adversarial Code Obfuscation Against CoT-Enhanced LLM Code Agents

Affects: DeepSeek R1, GPT-5

Source: arXiv

Familiar Pattern Analysis Hijack

Updated 2/22/2026

Large Language Models (LLMs) utilized for static code analysis, code review, and autonomous software engineering exhibit a cognitive vulnerability termed "Abstraction Bias." When processing code that structurally resembles common algorithmic patterns (e.g., standard sorting algorithms, helper functions, or mathematical formulas), the model relies on high-level memorized representations of the algorithm's intent rather than analyzing the specific local logic. Adversaries can exploit this by…

Trust Me, I Know This Function: Hijacking LLM Static Analysis using Bias

Affects: GPT-4o, Claude 3.5 Sonnet, Gemini 2.0 Flash +3 more

Source: arXiv

LLM Hate Campaign Vulnerability

Updated 2/2/2025

Large Language Models (LLMs) used in hate speech detection systems are vulnerable to adversarial attacks and model stealing, resulting in evasion of hate speech detection. Adversarial attacks modify hate speech text to evade detection, while model stealing creates surrogate models that mimic the target system's behavior.

HateBench: Benchmarking Hate Speech Detectors on LLM-Generated Content and Hate Campaigns

Affects: Baichuan 2, Dolly 2, GPT-3.5 Turbo +2 more

Source: arXiv

Context-Shifting Code Injection

Updated 12/29/2024

Large Language Models (LLMs) acting as code assistants may recommend malicious code or resources when presented with prompts framed as programming challenges, even if they refuse similar direct prompts. This occurs due to insufficient context-aware safety mechanisms. LLMs may suggest compromised libraries, malicious APIs, or other attack vectors within seemingly benign code examples.

Hallucinating AI Hijacking Attack: Large Language Models and Malicious Code Recommenders

Affects: GPT-4

Source: arXiv

LLM Robot Bias & Violence

Updated 4/12/2025

Large Language Models (LLMs) used to control robots exhibit biases leading to discriminatory and unsafe behaviors. When provided with personal characteristics (e.g., race, gender, disability), LLMs generate biased outputs resulting in discriminatory actions (e.g., assigning lower rescue priority to certain groups) and accept or deem feasible dangerous or unlawful instructions (e.g., removing a person's mobility aid).

Llm-driven robots risk enacting discrimination, violence, and unlawful actions

Affects: GPT-3.5, GPT-3.5 Turbo, GPT-4 +1 more

Source: arXiv

Related research categories