mlsanityData to 14 Oct 2026

OPBackdoor: Opportunistic Backdoors via Alibi-Aligned Reasoning

arXiv 2609.24826 ↗First posted 21 Sep 2026cs.CRcited by 0

On the map: LLM safety and security → Backdoor and data poisoning attacks

Abstract (arXiv)

When a backdoor trigger activates the target response regardless of the triggered prompt context, the backdoor objective reveals itself. Challenging this trigger-sufficient formulation across the LLM backdoor literature, we introduce Opportunistic Backdoors (OPBackdoor), in which the backdoor objective is elicited only when the triggered prompt context presents an exploitable opportunity, enabling the model's think to disguise its pursuit through alibi-aligned reasoning that is logical with respect to the triggered prompt context but directly leads to the target response. Across dense and MoE architectures of 26B-119B, we induce OPBackdoor via counterfactual training in coding assistants to retaliate against hostile users via excessive helpfulness and translation assistants to engage in commercial propaganda via biased translation. Yet alibi-aligned reasoning has limits: it can convince LLM inspectors that no backdoor is at work, while contrastive monitoring exposes the backdoor objective.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Backdoor and data poisoning attacks, one of 12 directions in LLM safety and security. Builds on 26 of its 43 references that are in the map. First posted on arXiv 21 Sep 2026.

Builds on

26 of 43 references.

Closest ancestors · 14

Same direction, newest first

  1. 2026Stealthy Backdoor Attacks against LLMs Based on Natural Style Triggers23 Apr
  2. 2026MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning8 Apr
  3. 2026The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers3 Feb
  4. 2025ShadowCoT: Cognitive Hijacking for Stealthy Reasoning Backdoors in LLMs8 Apr
  5. 2025DarkMind: Latent Chain-of-Thought Backdoor in Customized LLMs24 Jan
  6. 2024When Backdoors Speak: Understanding LLM Backdoor Attacks Through Model-Generated Explanations19 Nov
  7. 2024Claim-Guided Textual Backdoor Attack for Practical Applications25 Sep
  8. 2024SynGhost: Invisible and Universal Task-agnostic Backdoor Attack via Syntactic Transfer29 Feb
  9. 2024Universal Vulnerabilities in Large Language Models: Backdoor Attacks for In-context Learning11 Jan
  10. 2024Sleeper Agents: Training Deceptive LLMs that Persist Through Safety Training10 Jan
  11. 2023Composite Backdoor Attacks Against Large Language Models11 Oct
  12. 2023Backdoor Attacks for In-Context Learning with Language Models27 Jul
2 more
  1. 2021Mind the Style of Text! Adversarial and Backdoor Attacks Based on Text Style Transfer14 Oct
  2. 2021Turn the Combination Lock: Learnable Textual Backdoor Attacks via Word Substitution11 Jun

Other references · 12

  1. 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscationcited by 384
  2. 2020BadNL: Backdoor Attacks against NLP Models with Semantic-preserving Improvementscited by 211
  3. 2023Universal Jailbreak Backdoors from Poisoned Human Feedbackcited by 128
  4. 2025Stress Testing Deliberative Alignment for Anti-Scheming Trainingcited by 86
  5. 2025LeetCodeDataset: A Temporal Dataset for Robust Evaluation and Efficient Training of Code LLMscited by 70
  6. 2026On the Design of Qwen3.8-Next Architecture: Evaluation, Efficiency, and Training Stabilitycited by 35
  7. 2026Reasoning Models Struggle to Control their Chains of Thoughtcited by 25
  8. 2026Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillationcited by 8
  9. 2025DecepChain: Inducing Deceptive Reasoning in Large Language Modelscited by 3
  10. 2026FunPRM: Function-as-Step Process Reward Model with Meta Reward Correction for Code Generationcited by 3
  11. 2025GREAT: Generalizable Backdoor Attacks in RLHF via Emotion-Aware Trigger Synthesiscited by 1
  12. 2025SteganoBackdoor: Evading Data-Poisoning Defenses via Steganographic Backdoorscited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025From Poisoned to Aware: Fostering Backdoor Self-Awareness in LLMs
  2. 2026AKRASIA: Stealthy Backdoor Attack on Reasoning-based Code LLMs
  3. 2026Backdoor Unlearning Generalization: A Path Toward the Removal of Unknown Triggers in LLMs
  4. 2026Turn-Based Structural Triggers: Structure-Conditioned Backdoors in Multi-Turn LLMs
  5. 2025Rethinking Reasoning: A Survey on Reasoning-based Backdoors in LLMs
  6. 2025Thought Crime: Backdoors and Emergent Misalignment in Reasoning Models
  7. 2025SLIP: Soft Label Mechanism and Key-Extraction-Guided CoT-based Defense Against Instruction Backdoor in APIs
  8. 2026Purifying Generative LLMs from Backdoors without Prior Knowledge or Clean Reference