CacheSolidarity: Preventing Prefix Caching Side Channels in Multi-Tenant LLM Serving Systems
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Security Radio. Generated commentary on the latest security and cryptography papers.
Nadia: I'm Nadia.
Elias: And I'm Elias, here with Priya.
Nadia: We're looking at the paper: "CacheSolidarity: Preventing Prefix Caching Side Channels in Multi-Tenant LLM Serving Systems".
Elias: Here's the gist of the paper. Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference, but this introduces timing side channels that attackers can exploit to infer sensitive information in multi-tenant systems The paper presents PrefixWall, a system that secures multi-tenant LLM serving systems against APC side channels without sacrificing performance and efficiency Problem Statement The core issue is that APC creates observable latency differences between cache hits and misses, which form a timing side channel that attackers can use to reconstruct another user’s request by observing hit/miss patterns Current defenses take a "sledgehammer approach" by disabling APC and isolating users, which sacrifices efficiency for regular users Existing defenses suffer from significant overhead because they isolate entire users instead of just the shared prefixes that might lead to attacks PrefixWall Design PrefixWall is a system-level solution based on the insight that isolating entire users is not necessary, but rather sufficient to protect the shared prefixes that might lead to attacks The design involves three main components: 1.
Nadia: KV Cache Extension: Augments each cache entry with metadata like OwnerID and AttackFlag to track ownership and flag suspicious reuse 2.
Elias: Detector: Examines prefix reuse to identify cross-user access and enforces selective isolation when suspicious reuse occurs 3.
Nadia: Activator: Dynamically activates selective isolation only under conditions that create exploitable timing side-channels, balancing security with performance Performance and Evaluation Evaluation shows that PrefixWall enables up to 70% higher cache reuse and 30% lower inference latency compared to existing defenses that isolate users The system is implemented on top of vLLM
64: and demonstrates negligible time and memory overheads during runtime PrefixWall preserves most of the performance benefits of prefix caching while enforcing strong security guarantees, even under diverse workload patterns Security Guarantees PrefixWall's selective isolation scheme is designed to provide security against prompt-stealing attacks performed through APC timing side channels Security guarantees are provided when conditions (1) and (2) are met, ensuring that executing all probes in the sequence s from user identifiers different from u will always result in misses for entries v1,..., vn The system successfully closes the timing leak on an "attacker workload," where both hit rate and TTFT are uniform across all attacker requests Sensitivity Analysis The selection of the KDE overlap threshold represents a tradeoff between security and performance With very low thresholds, PrefixWall performs comparably to Prefix Caching for the same workload since selective-isolation is almost always disabled For high threshold (close to 100%), the system achieves the hit rate and performance of Selective Cache Isolation The activator component provides the system administrator with a way of effectively balancing performance and security Overhead Analysis In terms of per-request latency, PrefixWall’s modifications introduce an average per-request Component Time of 0.007 ms (0.004 ms due to the detector component and 0.003 ms due to the activator component) In terms of memory, the metadata introduced by the detector component amounts to 32 bytes per cache entry The overall overhead is minimal, demonstrating that security can be achieved without sacrificing performance when implemented in a lightweight and principled way PrefixWall builds on the key insight that preventing timing side-channel attacks does not require isolating entire users, but only requires isolating prefixes that might lead to attacks This approach enables efficient sharing for benign users while selectively isolating only those prefixes associated with suspicious behavior PrefixWall’s lightweight design demonstrates how security in LLM serving does not have to come at the cost of unnecessarily reduced performance or unbearable overheads The specific paper contributions include a detailed analysis of timing side channels, the design of PrefixWall, a comprehensive security analysis, and an extensive evaluation across diverse workloads Page 1 --- PrefixWall: Mitigating Prefix Caching Side Channels in Shared LLM Systems Panagiotis Georgios Pennas IMDEA Software Institute Universidad Politécnica de Madrid panagiotis.pennas@imdea.org Konstantinos Papaioannou IMDEA Software Institute Universidad Politécnica de Madrid konstatinos.papaioannou@imdea.org Marco Guarnieri IMDEA Software Institute marco.guarnieri@imdea.org Thaleia Dimitra Doudali IMDEA Software Institute thaleia.doudali@imdea.org ABSTRACT Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference APC works by reusing previously computed states for the beginning part of a request (prefix), when another request starts with the same text While APC improves throughput, it introduces timing side channels: cache hits are faster than misses, creating observable latency differences In multi-tenant systems, attackers can exploit these differences to infer sensitive information, e.g., by incrementally reconstructing another user’s request by observing hit/miss patterns Current defenses take a sledgehammer approach: they disable APC and cache sharing, isolating users, and sacrificing efficiency for regular users This paper presents PrefixWall, a system that secures multi-tenant LLM serving systems against APC side channels without sacrificing performance and efficiency PrefixWall monitors cache reuse across users, flags suspicious sharing, and selectively isolates prefixes, restricting their reuse, only when necessary Evaluation shows that PrefixWall enables up to 70% higher cache reuse and 30% lower inference latency compared to existing defenses that isolate users PrefixWall’s lightweight design demonstrates how security in LLM serving does not have to come at the cost of unnecessarily reduced performance or unbearable overheads 1 INTRODUCTION Large Language Models (LLMs) now power applications such as conversational assistants, code generation, and enterprise analytics
13: These services operate at massive scale under strict latency and throughput requirements, making inference serving a critical systems challenge
76: Modern serving systems employ system-level optimizations such as cache management
39, 40, 59: and scheduling
2, 3, 33: to accelerate LLM inference Among these optimizations, prefix sharing
10, 37, 71: , or otherwise known as Automatic Prefix Caching (APC), is widely used by state-of-the-art frameworks and commercial APIs APC accelerates inference by caching and reusing previously computed model states for the beginning part of a request (the request’s prefix) whenever another request starts with the same text The use of a prefix cache avoids redundant computation and significantly reduces latency for long prompts and multi-turn conversations
37, 71: While APC delivers substantial performance benefits, it also introduces security risks similar to other caching optimizations [12, Name: Password: Address: Email: Prompt Constructor Hit Miss FEEDBACK Shared Prefix Cache Time Analyzer Victim Attacker Serving System LLM Secret Inputs Request Response Request Response … Response Generation Time Request Diff Response LLM Figure 1 Timing side-channel leakage in prefix-sharing LLM inference.
Paper summary: Elias: So to wrap up the discussion on "CacheSolidarity: Preventing Prefix Caching Side Channels in Multi-Tenant LLM Serving Systems", the authors are really pushing the idea that you don't need full user isolation to stop prompt-stealing attacks exploiting APC timing side channels.
Priya: It’s about moving away from that sledgehammer approach and toward a more surgical security method where you only isolate specific prefixes when there's actual suspicious behavior, which they achieve with the PrefixWall system architecture.
Nadia: Right. The implication is that for high-performance, multi-tenant LLM serving environments, this selective isolation can be the right balance to maintain efficiency while closing those timing leaks without imposing massive penalties on benign users.
Elias: They show that even with their lightweight design, they can achieve up to seventy percent higher cache reuse and lower latency compared to defenses that enforce user-level isolation <ref:2603.10726#pg1>.
Priya: So, the takeaway for us is that security in this area can be achieved through a principled design focused on monitoring prefix reuse across users and selectively isolating only those prefixes associated with suspicious behavior, as outlined in "CacheSolidarity: Preventing Prefix Caching Side Channels in Multi-Tenant LLM Serving Systems".
Conclusion: Nadia: So, we're wrapping up this look at "CacheSolidarity: Preventing Prefix Caching Side Channels in Multi-Tenant LLM Serving Systems". The main idea is that you don't need to isolate every single user completely to stop prompt-stealing attacks when using Automatic Prefix Caching.
Elias: Right. They’re arguing that isolating entire users is too much overhead for regular users, and the actual security comes from selectively isolating just the shared prefixes when they look suspicious across different users.
Priya: From a measurement standpoint, what this means is that instead of crippling performance with heavy isolation, you can maintain pretty good cache reuse while still mitigating that timing leak risk.
Nadia: Exactly. The authors show that this selective approach actually allows for up to seventy percent higher cache reuse and thirty percent lower inference latency compared to the other defenses they tested.
Elias: I like the numbers, but we gotta look at how they do it technically, you know? They extend the cache entry with some metadata like an Owner ID and an Attack Flag.
Priya: That metadata is key. It lets them track ownership and flag reuse patterns between users, which feeds into a detector that isolates prefixes only when things get too suspicious.
Nadia: The system has three parts: the cache extension, the detector that flags suspicious sharing, and then the activator that decides when to actually enforce isolation based on whether a timing side channel is exploitable.
Elias: And they validate it with a security analysis showing that under certain conditions—specifically when certain probes are made from different user identifiers—the system ensures misses for those entries.
Priya: The sensitivity analysis on that overlap threshold shows the trade-off between performance and security, letting the admin fine-tune how strict the isolation gets.
Nadia: So, what does this mean for real world deployment? It suggests that securing LLM serving doesn't always have to mean sacrificing speed for every single request.
Elias: It means finding a middle ground where you get strong protection against these specific timing attacks without making the system unusable for everyone who just wants fast inference.
Priya: If this holds up under diverse workloads, it could be a significant way to make shared LLM infrastructure safer in practice.
IMDEA Software Institute · Universidad Politécnica de Madrid
cs.CR, cs.DC, cs.LG
Submitted: 2026-03-11
Updated: 2026-10-29
Code: https://github.com/vusec/revancPress=https:
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 92/100
The gist: Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference…
Key concepts
- Automatic Prefix Caching (APC)
- APC is an optimization used in LLM serving that reuses previously computed model states for the beginning part of a request, called a prefix. While it speeds up inference by avoiding redundant computation, it creates observable latency differences between cache hits and misses that attackers can exploit.
- Timing Side Channel
- This is an unintended information leak where the time taken to process a request reveals properties of secret data. In this context, APC causes timing variations based on whether a prefix is cached or not, allowing an attacker to infer parts of another user's sensitive prompt by measuring these small latency differences.
- PrefixWall
- A system designed to prevent APC side channels by monitoring cache reuse across users. Instead of disabling caching entirely, it tracks suspicious sharing and selectively isolates only the prefixes involved in potential attacks, balancing security with high performance.
Terminology
Summary
Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference, but this introduces timing side channels that attackers can exploit to infer sensitive information in multi-tenant systems The paper presents PrefixWall, a system that secures multi-tenant LLM serving systems against APC side channels without sacrificing performance and efficiency
Problem Statement
The core issue is that APC creates observable latency differences between cache hits and misses, which form a timing side channel that attackers can use to reconstruct another user’s request by observing hit/miss patterns Current defenses take a sledgehammer approach
by disabling APC and isolating users, which sacrifices efficiency for regular users Existing defenses suffer from significant overhead because they isolate entire users instead of just the shared prefixes that might lead to attacks
PrefixWall Design
PrefixWall is a system-level solution based on the insight that isolating entire users is not necessary, but rather sufficient to protect the shared prefixes that might lead to attacks The design involves three main components:
-
KV Cache Extension: Augments each cache entry with metadata like OwnerID and AttackFlag to track ownership and flag suspicious reuse
-
Detector: Examines prefix reuse to identify cross-user access and enforces selective isolation when suspicious reuse occurs
-
Activator: Dynamically activates selective isolation only under conditions that create exploitable timing side-channels, balancing security with performance
Performance and Evaluation
Evaluation shows that PrefixWall enables up to 70% higher cache reuse and 30% lower inference latency compared to existing defenses that isolate users The system is implemented on top of vLLM [64] and demonstrates negligible time and memory overheads during runtime PrefixWall preserves most of the performance benefits of prefix caching while enforcing strong security guarantees, even under diverse workload patterns
Security Guarantees
PrefixWall's selective isolation scheme is designed to provide security against prompt-stealing attacks performed through APC timing side channels Security guarantees are provided when conditions (1) and (2) are met, ensuring that executing all probes in the sequence s from user identifiers different from u will always result in misses for entries v1,..., vn The system successfully closes the timing leak on an attacker workload,
where both hit rate and TTFT are uniform across all attacker requests
Sensitivity Analysis
The selection of the KDE overlap threshold represents a tradeoff between security and performance With very low thresholds, PrefixWall performs comparably to Prefix Caching for the same workload since selective-isolation is almost always disabled For high threshold (close to 100%), the system achieves the hit rate and performance of Selective Cache Isolation The activator component provides the system administrator with a way of effectively balancing performance and security
Overhead Analysis
In terms of per-request latency, PrefixWall’s modifications introduce an average per-request Component Time of 0.007 ms (0.004 ms due to the detector component and 0. p. 8 Note
also that, for all experiments in 5. 2, the detector component of
PrefixWall is always active.Where the paper says it
Conclusion
PrefixWall builds on the key insight that preventing timing side-channel attacks does not require isolating entire users, but only requires isolating prefixes that might lead to attacks This approach enables efficient sharing for benign users while selectively isolating only those prefixes associated with suspicious behavior PrefixWall’s lightweight design demonstrates how security in LLM serving does not have to come at the cost of unnecessarily reduced performance or unbearable overheads The specific paper contributions include a detailed analysis of timing side channels, the design of PrefixWall, a comprehensive security analysis, and an extensive evaluation across diverse workloads
--- Page 1 ---
PrefixWall: Mitigating Prefix Caching Side Channels in Shared LLM Systems Panagiotis Georgios Pennas IMDEA Software Institute Universidad Politécnica de Madrid panagiotis.pennas@imdea. p. 0 PrefixWall: Mitigating Prefix Caching Side Channels in Shared
LLM Systems
Panagiotis Georgios Pennas
IMDEA Software Institute
Universidad Politécnica de Madrid
panagiotis. pennas@imdea. pennas@imdea. org
Konstantinos Papaioannou
IMDEA Software Institute
Universidad Politécnica de Madrid
konstantinos. pennas@imdea. org
Konstantinos Papaioannou
IMDEA Software Institute
Universidad Politécnica de Madrid
konstantinos. papaioannou@imdea. org
Marco Guarnieri
IMDEA Software Institute
marco. papaioannou@imdea. org
Marco Guarnieri
IMDEA Software Institute
marco. guarnieri@imdea. org
Thaleia Dimitra Doudali
IMDEA Software Institute
thaleia. guarnieri@imdea. org
Thaleia Dimitra Doudali
IMDEA Software Institute
thaleia. org
ABSTRACT
Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference. APC works
by reusing previously computed states for the beginning part of
a request (prefix), when another request starts with the same text. The
use of a prefix cache avoids redundant computation and significantly reduces latency for long prompts and multi-turn conversations [37, 71]. While APC delivers substantial performance benefits, it also introduces security risks similar to other caching optimizations [12,
Name: **
Password: **
Address: **
Email: **
Prompt
Constructor
Hit
Miss
FEEDBACK
Shared
Prefix
Cache
Time
Analyzer
Victim
Attacker
Serving System
LLM
Secret Inputs
Request
Response
Request
Response
…
Response
Generation
Time
Request Diff
Response
LLM Figure 1: Timing side-channel leakage in prefix-sharing LLM
inference. APC creates observable variations in inference latency
depending on whether a request reuses cached prefixes or not. These latency differences create a timing side channel, that is, an unintended information leak where execution time reveals properties
of secret data without direct access.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
--- Page 2 ---
thereby limiting their applicability in realistic, high-performance multi-tenant environments To address the shortcomings of current defenses, we introduce PrefixWall, a system-level solution that secures multi-tenant LLM serving against timing side channels caused by APC, while preserving its performance benefits PrefixWall is based on the key insight that isolating entire users (as done in [31, 51, 75]) is not necessary to prevent attacks; rather it is sufficient to protect the shared prefixes that might lead to attacks Building on this insight, PrefixWall (1) continuously monitors prefix reuse across users, (2) flags prefixes that are suspiciously reused by multiple users, and (3) prevents further reuse of flagged prefixes in follow-up requests This approach enables efficient sharing for benign users, while selectively isolating only those prefixes associated with suspicious behavior, avoiding unnecessary penalties for all users The system design of PrefixWall consists of lightweight components that work together to secure prefix caching without sacrificing efficiency First, PrefixWall extends the prefix cache with minimal metadata for each entry to track ownership and flag suspicious reuse Second, the Detector monitors cache hits and isolates flagged prefixes when multiple users attempt to share them Finally, PrefixWall introduces a system-level optimization that activates prefix isolation only when a timing side channel is actually exploitable This is handled by the Activator, which continuously evaluates whether latency differences between cache hits and misses are distinguishable, based on factors such as the hardware platform, LLM model size, system load, and request length Overall PrefixWall’s system components interact seamlessly: the cache extension provides metadata, the Detector enforces selective isolation, and the Activator optimizes the security-performance trade-off PrefixWall is implemented on top of the open-source, state-of-the-art LLM serving system vLLM [64] We conduct an extensive evaluation across multi-tenant workloads with varying levels of intra- and inter-user prefix sharing and nine LLM models spanning different families and sizes (from 0.5B to 13B parameters) Our experimental results show that PrefixWall achieves up to 70% higher cache reuse and 30% lower inference latency compared to defenses that enforce user-level isolation In addition, PrefixWall introduces negligible time and memory overheads during runtime, demonstrating that security can be achieved without sacrificing performance when implemented in a lightweight and principled way Finally, to validate the security of PrefixWall, we performed a detailed security analysis that precisely characterizes PrefixWall’s security guarantees, which we complement with an empirical validation showing that PrefixWall indeed closes the timing side-channel introduced by APC across requests from different users The specific paper contributions are: A detailed analysis of timing side channels introduced by Automatic Prefix Caching (APC) in multi-tenant LLM serving systems, including observations on their exploitability under different workload and system conditions (Section 2) The design of PrefixWall, a lightweight and practical system that secures prefix reuse against timing attacks by selectively isolating suspicious prefixes rather than entire users, while preserving performance benefits (Section 3) A comprehensive security analysis of PrefixWall’s guarantees and limitations, complemented by empirical validation against prompt-stealing attacks (Section 4) An extensive evaluation of PrefixWall across diverse workloads and state-of-the-art LLM models, demonstrating up to 70% higher cache reuse and 30% lower latency compared to user-level isolation defenses, with negligible overhead (Section 5)
--- Page 3 ---
In this section, we provide background and motivation for PrefixWall We start by introducing background on LLM serving systems and, in particular, on the APC optimization (2.1) We then characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2. p. 1 We start by introducing background on LLM serving systems and, in particular, on the APC optimization (2. We then
characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2. We start by introducing background on LLM serving systems and, in particular, on the APC optimization (2. We then
characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2. We then
characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2. Then, we illustrate
how attackers can exploit APC through timing leaks to learn sensitive prompts (2. We then
characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2. Then, we illustrate
how attackers can exploit APC through timing leaks to learn sensitive prompts (2. Then, we illustrate
how attackers can exploit APC through timing leaks to learn sensitive prompts (2. Next, we overview current defenses for APC
timing leaks and discuss their limitations (2. Then, we illustrate
how attackers can exploit APC through timing leaks to learn sensitive prompts (2. Next, we overview current defenses for APC
timing leaks and discuss their limitations (2. Next, we overview current defenses for APC
timing leaks and discuss their limitations (2. We conclude by
summarizing our motivational observations in (2. Next, we overview current defenses for APC
timing leaks and discuss their limitations (2. We conclude by
summarizing our motivational observations in (2.Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
--- Page 4 ---
In this section, we provide background and motivation for PrefixWall We start by introducing background on LLM serving systems and, in particular, on the APC optimization (2.1) We then characterize the parameters that influence the exploitability of timing side channels due to APC (Section 2.2) Then, we illustrate how attackers can exploit APC through timing leaks to learn sensitive prompts (2.3) Next, we overview current defenses for APC timing leaks and discuss their limitations (2.4) We conclude by summarizing our motivational observations in (2.
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Secure prefix reuse by isolating suspicious prefixes instead of entire users, which allows
benign users to benefit from shared caching
while stopping reuse beyond flagged entries for non-owners. -
Dynamically activate selective isolation based on whether timing signals are distinguishable, ensuring that
PrefixWall deactivates selective isolation and allows full prefix reuse
when the KDE overlap exceeds a threshold of theta. -
Prevent prompt stealing attacks by ensuring that
executing all probes in the sequence s from user identifiers different from u will always result in misses for entries v1, . . . , vn
under specific conditions. -
Preserve performance by ensuring
PrefixWall preserves most of the performance benefits of prefix caching while enforcing strong security guarantees,
maintaining a TTFT and cache hit rate close to the unprotected Prefix Caching baseline across diverse workloads.
Abstract
Large Language Models (LLMs) rely on optimizations like Automatic Prefix Caching (APC) to accelerate inference. APC works by reusing previously computed states for the beginning part of a request (prefix), when another request starts with the same text. While APC improves throughput, it introduces timing side channels: cache hits are faster than misses, creating observable latency differences. In multi-tenant systems, attackers can exploit these differences to infer sensitive information, e.g., by incrementally reconstructing another user's request by observing hit/miss patterns. Current defenses take a sledgehammer approach: they disable APC and cache sharing, isolating users, and sacrificing efficiency for regular users. This paper presents CacheSolidarity, a system that secures multi-tenant LLM serving systems against APC side channels without sacrificing performance and efficiency. CacheSolidarity monitors cache reuse across users, flags suspicious sharing, and selectively isolates prefixes, restricting their reuse only when necessary. Evaluation shows that CacheSolidarity enables up to 70% higher cache reuse and 30% lower inference latency compared to existing user-isolation defenses. CacheSolidarity's lightweight design demonstrates how LLM serving can be secured without sacrificing performance or incurring unbearable overheads.
Sources
- No Request Left Behind: Tackling Heterogeneity in Long-Context LLM Inference with Medha
- SARATHI: Efficient LLM Inference by Piggybacking Decodes with Chunked Prefills
- Remote Timing Attacks on Efficient Language Model Inference
- Efficient Heterogeneous Large Language Model Decoding with Model-Attention Disaggregation
- Selective KV-Cache Sharing to Mitigate Timing Side-Channels in LLM Inference
- HashAttention: Semantic Sparsity for Faster Inference
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- LazyLLM: Dynamic Token Pruning for Efficient Long Context LLM Inference
- Prompt Cache: Modular Attention Reuse for Low-Latency Inference
- Auditing Prompt Caching in Language Model APIs
- DeepSpeed-FastGen: High-throughput Text Generation for LLMs via MII and DeepSpeed-Inference
- MInference 1.0: Accelerating Pre-filling for Long-Context LLMs via Dynamic Sparse Attention
- Hydragen: High-Throughput LLM Inference with Shared Prefixes
- Ring Attention with Blockwise Transformers for Near-Infinite Context
- The Early Bird Catches the Leak: Unveiling Timing Side Channels in LLM Serving Systems
- ShadowKV: KV Cache in Shadows for High-Throughput Long-Context LLM Inference
- LLaMA: Open and Efficient Foundation Language Models
- Efficient Streaming Language Models with Attention Sinks
- Prepacking: A Simple Method for Fast Prefilling and Increased Throughput in Large Language Models
- InputSnatch: Stealing Input in LLM Services via Timing Side-Channel Attacks
Related papers
- From Strategy Search to Policy Learning: MCTS-Guided Adversarial Co-Evolution for LLM Safety Alignment
- Privacy Leakage from Recurrent State in LLM Inference
- Incrementally Verifiable Oracle Aided Computation
- Privilege and confidentiality in generative AI workflows
- BadMoE: Backdooring Mixture-of-Experts LLMs via Routing-Aware Triggers and Task-Underutilized Experts
- CoVault: Secure, Scalable Analytics of Personal Data