What the Middleman Delivers: Answer Behavior, Tool Permissions, and Billed Usage in LLM API Relays
cs.SE, cs.AI
Submitted: 2026-11-05
Updated: 2026-11-05
Code: https://github.com/toby-bridges/apirelay-audit
Terminology
Sources
- Real Money, Fake Models: Deceptive Model Claims in Shadow APIs
- The Proxy Knows Too Much: Sealing LLM API Routers with Attested TEEs
- KBF: Knowledge Boundary as Fingerprint for Language Model and Black-Box API Auditing
- Confidential Prompting: Privacy-preserving LLM Inference on Cloud
- Mind your key: An Empirical Study of LLM API Credential Leakage in iOS Apps
- One Token Is Enough: Fingerprinting and Verifying Large Language Models from Single-Token Output Distributions
- SVIP: Towards Verifiable Inference of Open-source Large Language Models
- Privacy-Preserving Mechanisms Enable Cheap Verifiable Inference of LLMs
- Your Agent Is Mine: Measuring Malicious Intermediary Attacks on the LLM Supply Chain
- Which Model Is Actually Serving You? IRIS: Budgeted Black-Box Auditing of Model Substitution and Routing Dilution in LLM Gateways
- Are You Getting What You Pay For? Auditing Model Substitution in LLM APIs
- DiFR: Inference Verification Despite Nondeterminism
- Your "Pro" LLM Subscription May Actually Be "Free": Exposing Fingerprint Spoofing Risks in LLM Inference Services
- Rewriting the Response Path: Silent Tampering and Provider-Signed Defense in BYOK LLM Agents
- AEX: Non-Intrusive Multi-Hop Attestation and Provenance for LLM APIs
- Evidence-Bound Gateway-Path Provenance for Third-Party LLM Inference
- Injecting Falsehoods: Adversarial Man-in-the-Middle Attacks Undermining Factual Recall in LLMs
- Attacks on Third-Party APIs of Large Language Models
Related papers
- ArkEval: Benchmarking and Evaluating Automated CodeRepair for ArkTS
- TestPrism: Rethinking Test Evaluation Beyond a Single Reference
- Mining Workflow Graphs for Black-Box Boundary Testing of Conversational LLM Agents
- Falsification-Based Verification of LLM-Generated Optimization Models: Sound Test Batteries and Their Detection Limits
- GitSkills: A Dataset of Agent Skills on GitHub
- For Coding Agents, the Repository Matters