# Get my drift? Catching LLM Task Drift with Activation Deltas

arXiv: https://arxiv.org/abs/2406.00799 · HTML: https://mlsanity.com/paper/2406.00799/

## Key facts

- Cited by 72 papers in the map (data through 14 Oct 2026).
- Placed in Privacy and security of RAG, one of 15 directions in Retrieval-augmented generation.
- Builds on 39 of its 56 references that are in the map.
- First posted on arXiv 2 Jun 2024.

## Placed in

- Retrieval-augmented generation / [Privacy and security of RAG](https://mlsanity.com/rag/privacy-and-security-of-rag/)
- LLM safety and security / [Prompt injection and agent security](https://mlsanity.com/safety/prompt-injection-and-agent-security/)

## Builds on (most cited first, 30 of 39)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [SQuAD: 100,000+ Questions for Machine Comprehension of Text](https://mlsanity.com/paper/1606.05250/) (2016)
- [HotpotQA: A Dataset for Diverse, Explainable Multi-hop Question Answering](https://mlsanity.com/paper/1809.09600/) (2018)
- [Mistral 7B](https://mlsanity.com/paper/2310.06825/) (2023)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [FaceNet: A Unified Embedding for Face Recognition and Clustering](https://mlsanity.com/paper/1503.03832/) (2015)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [Not what you've signed up for: Compromising Real-World LLM-Integrated Applications with Indirect Prompt Injection](https://mlsanity.com/paper/2302.12173/) (2023)
- [Jailbreaking Black Box Large Language Models in Twenty Queries](https://mlsanity.com/paper/2310.08419/) (2023)
- [Representation Engineering: A Top-Down Approach to AI Transparency](https://mlsanity.com/paper/2310.01405/) (2023)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [Discovering Latent Knowledge in Language Models Without Supervision](https://mlsanity.com/paper/2212.03827/) (2022)
- [WildChat: 1M ChatGPT Interaction Logs in the Wild](https://mlsanity.com/paper/2405.01470/) (2024)
- [Eliciting Latent Predictions from Transformers with the Tuned Lens](https://mlsanity.com/paper/2303.08112/) (2023)
- [JailbreakBench: An Open Robustness Benchmark for Jailbreaking Large Language Models](https://mlsanity.com/paper/2404.01318/) (2024)
- [The Instruction Hierarchy: Training LLMs to Prioritize Privileged Instructions](https://mlsanity.com/paper/2404.13208/) (2024)
- ["Do Anything Now": Characterizing and Evaluating In-The-Wild Jailbreak Prompts on Large Language Models](https://mlsanity.com/paper/2308.03825/) (2023)
- [Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks](https://mlsanity.com/paper/2404.02151/) (2024)
- [StruQ: Defending Against Prompt Injection with Structured Queries](https://mlsanity.com/paper/2402.06363/) (2024)
- [In-Context Learning Creates Task Vectors](https://mlsanity.com/paper/2310.15916/) (2023)
- [Low-Resource Languages Jailbreak GPT-4](https://mlsanity.com/paper/2310.02446/) (2023)
- [TrustLLM: Trustworthiness in Large Language Models](https://mlsanity.com/paper/2401.05561/) (2024)
- [PoisonedRAG: Knowledge Corruption Attacks to Retrieval-Augmented Generation of Large Language Models](https://mlsanity.com/paper/2402.07867/) (2024)
- [Defending Against Indirect Prompt Injection Attacks With Spotlighting](https://mlsanity.com/paper/2403.14720/) (2024)
- [Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models](https://mlsanity.com/paper/2312.14197/) (2023)
- [A Wolf in Sheep's Clothing: Generalized Nested Jailbreak Prompts can Fool Large Language Models Easily](https://mlsanity.com/paper/2311.08268/) (2023)
- [Jatmo: Prompt Injection Defense by Task-Specific Finetuning](https://mlsanity.com/paper/2312.17673/) (2023)
- [JailbreakRadar: Comprehensive Assessment of Jailbreak Attacks Against LLMs](https://mlsanity.com/paper/2402.05668/) (2024)

## Built on by (most cited first, 30 of 72)

- [Advances and Challenges in Foundation Agents: From Brain-Inspired Intelligence to Evolutionary, Collaborative, and Safe Systems](https://mlsanity.com/paper/2504.01990/) (2025)
- [Securing AI Agents with Information-Flow Control](https://mlsanity.com/paper/2505.23643/) (2025)
- [Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks](https://mlsanity.com/paper/2602.20156/) (2026)
- [Design Patterns for Securing LLM Agents against Prompt Injections](https://mlsanity.com/paper/2506.08837/) (2025)
- [Navigating the Risks: A Survey of Security, Privacy, and Ethics Threats in LLM-Based Agents](https://mlsanity.com/paper/2411.09523/) (2024)
- [AttriGuard: Defeating Indirect Prompt Injection in LLM Agents via Causal Attribution of Tool Invocations](https://mlsanity.com/paper/2603.10749/) (2026)
- [Firewalls to Secure Dynamic LLM Agentic Networks](https://mlsanity.com/paper/2502.01822/) (2025)
- [PromptLocate: Localizing Prompt Injection Attacks](https://mlsanity.com/paper/2510.12252/) (2025)
- [Agentic AI Security: Threats, Defenses, Evaluation, and Open Challenges](https://mlsanity.com/paper/2510.23883/) (2025)
- [Multi-Agent Large Language Models for Conversational Task-Solving](https://mlsanity.com/paper/2410.22932/) (2024)
- [ASIDE: Architectural Separation of Instructions and Data in Language Models](https://mlsanity.com/paper/2503.10566/) (2025)
- [LLMail-Inject: A Dataset from a Realistic Adaptive Prompt Injection Challenge](https://mlsanity.com/paper/2506.09956/) (2025)
- [Dataset and Lessons Learned from the 2024 SaTML LLM Capture-the-Flag Competition](https://mlsanity.com/paper/2406.07954/) (2024)
- [PIShield: Detecting Prompt Injection Attacks via Intrinsic LLM Features](https://mlsanity.com/paper/2510.14005/) (2025)
- [Depth Gives a False Sense of Privacy: LLM Internal States Inversion](https://mlsanity.com/paper/2507.16372/) (2025)
- [Overcoming the Retrieval Barrier: Indirect Prompt Injection in the Wild for LLM Systems](https://mlsanity.com/paper/2601.07072/) (2026)
- [PISmith: Reinforcement Learning-based Red Teaming for Prompt Injection Defenses](https://mlsanity.com/paper/2603.13026/) (2026)
- [A Framework for Formalizing LLM Agent Security](https://mlsanity.com/paper/2603.19469/) (2026)
- [Agent-Sentry: Bounding LLM Agents via Execution Provenance](https://mlsanity.com/paper/2603.22868/) (2026)
- [When Benchmarks Lie: Evaluating Malicious Prompt Classifiers Under True Distribution Shift](https://mlsanity.com/paper/2602.14161/) (2026)
- [LUMIA: Linear probing for Unimodal and MultiModal Membership Inference Attacks leveraging internal LLM states](https://mlsanity.com/paper/2411.19876/) (2024)
- [Meta SecAlign: A Secure Foundation LLM Against Prompt Injection Attacks](https://mlsanity.com/paper/2507.02735/) (2025)
- [Attention is All You Need to Defend Against Indirect Prompt Injection Attacks in LLMs](https://mlsanity.com/paper/2512.08417/) (2025)
- [Optimizing Agent Planning for Security and Autonomy](https://mlsanity.com/paper/2602.11416/) (2026)
- [Mitigating Indirect Prompt Injection via Instruction-Following Intent Analysis](https://mlsanity.com/paper/2512.00966/) (2025)
- [The Promptware Kill Chain: How Prompt Injections Gradually Evolved Into a Multistep Malware Delivery Mechanism](https://mlsanity.com/paper/2601.09625/) (2026)
- [Recent advancements in LLM Red-Teaming: Techniques, Defenses, and Ethical Considerations](https://mlsanity.com/paper/2410.09097/) (2024)
- [Security Considerations for Artificial Intelligence Agents](https://mlsanity.com/paper/2603.12230/) (2026)
- [Detecting Malicious Agent Skills in the Wild using Attention](https://mlsanity.com/paper/2606.23416/) (2026)
- [ARGUS: Defending Against Multimodal Indirect Prompt Injection via Steering Instruction-Following Behavior](https://mlsanity.com/paper/2512.05745/) (2025)

## Abstract (arXiv)

LLMs are commonly used in retrieval-augmented applications to execute user instructions based on data from external sources. For example, modern search engines use LLMs to answer queries based on relevant search results; email plugins summarize emails by processing their content through an LLM. However, the potentially untrusted provenance of these data sources can lead to prompt injection attacks, where the LLM is manipulated by natural language instructions embedded in the external data, causing it to deviate from the user's original instruction(s). We define this deviation as task drift. Task drift is a significant concern as it allows attackers to exfiltrate data or influence the LLM's output for other users. We study LLM activations as a solution to detect task drift, showing that activation deltas - the difference in activations before and after processing external data - are strongly correlated with this phenomenon. Through two probing methods, we demonstrate that a simple linear classifier can detect drift with near-perfect ROC AUC on an out-of-distribution test set. We evaluate these methods by making minimal assumptions about how users' tasks, system prompts, and attacks can be phrased. We observe that this approach generalizes surprisingly well to unseen task domains, such as prompt injections, jailbreaks, and malicious instructions, without being trained on any of these attacks. Interestingly, the fact that this solution does not require any modifications to the LLM (e.g., fine-tuning), as well as its compatibility with existing meta-prompting solutions, makes it cost-efficient and easy to deploy. To encourage further research on activation-based task inspection, decoding, and interpretability, we release our large-scale TaskTracker toolkit, featuring a dataset of over 500K instances, representations from six SoTA language models, and a suite of inspection tools.