Papers for

api security teams

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

API TAIL detects hidden model swaps in language model services

API-TAIL: Detecting Model Substitution in LLM APIs through Temperature-Aligned Likelihood Tails

Abstract: Commercial language model APIs return only text, allowing a provider to cut costs by quietly routing a fraction of user requests to a cheaper substitute model while billing for the advertised one. Detecting such partial substitution is challenging because honest providers also alter the output distribution through undocumented decoding and serving choices, such as temperature scaling, sampling truncation, repetition and presence penalties, and default system prompts, forcing an audit to test against a composite null of legitimate configurations rather than a single reference. Our key insight is that these legitimate variations primarily shift response typicality across the entire batch, whereas partial substitution injects individual responses that remain anomalous under the claimed model. Building on this, we introduce API-TAIL, a typicality audit that first absorbs batch-level shifts by aligning a single effective temperature to center standardized surprisal scores across the audited batch, and then flags the endpoint if the number of remaining tail responses exceeds a threshold calibrated across legitimate configurations. Evaluated on real-world commercial APIs, API-TAIL reliably controls false alarms across diverse honest providers while detecting even low-rate partial substitution with a small query budget, substantially outperforming existing rank- and MMD-based audits that lose detection power under the same composite calibration.

Tue 27 OctMachine LearningComputation and Language
The gist
Many companies offer language model services but might secretly use cheaper, different models for some user requests while charging for the original, making it hard to tell if the service is real. The authors developed a way to spot these hidden swaps by looking at unusual response patterns that don't match the claimed model, even when providers change settings like how creative or random responses are. Their method adjusts for normal shifts across a batch of answers, then checks for too many strange individual responses that could reveal substitution. Tests on real commercial APIs show this method detects hidden model swaps with few queries and fewer false alarms than older techniques.
Open → 2610.36801v1

Probeid verifies large language model tools despite changing responses

ProbeID: Verifying MCP Tool Identity for LLM Agents under Registry Drift

Abstract: Agents select tools by copyable descriptions, which cannot establish which implementation runs. ProbeID compares canonicalized responses to schema-valid probes with a trusted admission reference and abstains on ambiguity, so a squatter must copy probe-observable behavior. With the gold present, independent probes and a fixed margin, standard concentration arguments give a shared probe budget logarithmic in the candidate count and bound the joint probability of a wrong decision. On cached MCP-IdentityDrift cases, strict choice reaches \(0.956\) hit@1 while description routers fall near chance, but simpler exact comparisons tie it, MMD better tolerates heavy value noise, and a 32B behavioral judge is similarly accurate at higher cost. Random probes from agent banks and a prospective held-out study of 38 tools test theory-set thresholds. On the agent banks, equality refuses every reachable replacement from 64 probes and the bounded distance from 512, and sharper post hoc gates halve both. Two tools whose drift no bank probe reaches always pass. In an MCP-Bench agent, verification cuts tainted tasks from 15 to 9 of 16 at unchanged task success. Over eleven weeks, 7 of 59 real tools changed responses despite version pins. Verification depends on reference integrity, probe coverage, and response noise.

Sat 24 OctCryptography and SecurityArtificial Intelligence
The gist
Large language model (LLM) agents often pick tools to use based on descriptions that anyone can copy, making it hard to know which tool is really running. The authors propose ProbeID, a method that sends specific test questions called probes to tools and checks their responses against a trusted reference to confirm the tool's true identity. If the answers are ambiguous, ProbeID rejects the tool, so impostors must mimic observable behaviors exactly to pass. Tests show ProbeID can detect changes in tool behavior over time, improving task reliability by filtering out altered tools without hurting performance. The method depends on good reference data and coverage of probe questions.
Open → 2610.33633v1

Statistical auditing method measures identity risk of released text

Conformal Privacy Auditing: Calibrated Re-identification Attacks with Statistical Guarantees

Abstract: Empirical identity leakage from released text is increasingly driven by attackers that combine large language models (LLMs) with auxiliary knowledge to link documents to individuals. Existing audits typically report success rates for specific attack pipelines but lack finite-sample statistical guarantees, while training-time protections such as differential privacy are difficult to translate into release-time decisions for individual natural-language documents. We introduce Conformal Privacy Auditing(CPA), a distribution-free calibration framework that provides a statistical certificate of re-identification risk for each released document against LLM-empowered adversaries. CPA outputs a conformal ambiguity set of candidate identities that is guaranteed to contain the true identity with user-chosen confidence under exchangeability, together with an interpretable leakage proxy derived from set size. CPA supports both logit-access and sampling-only attackers, enabling audits of open-source models and proprietary API models in a unified framework. Across multiple release benchmarks and attacker configurations, CPA achieves calibrated coverage and reveals sharp shifts in certified identifiability as auxiliary knowledge, LLM augmentation, and release mechanisms vary, providing a statistically grounded basis for reporting and comparing release-time linkage risk across attacker configurations, datasets, and release mechanisms alike.

Fri 18 SeptCryptography and SecurityComputation and Language
The gist
Text documents released online can reveal who wrote them when attackers use powerful language AI and extra information to guess identities. The authors created a method called Conformal Privacy Auditing that gives a statistical guarantee on whether an identity guess is correct for each document. This method works with different kinds of AI models and helps figure out how risky it is to share certain texts. It shows when the chance of being identified goes up or down depending on how the text is prepared and what extra information attackers have.
Open → 2609.21340v1

Provider-side attacks inflate large language model outputs and costs

The More It Says, the More You Pay: A Black-Box Audit of Provider-Side Token Inflation in LLM Services

Abstract: In pay-per-token LLM services, the more a model says, the more users pay. Dishonest providers can covertly manipulate generation to inflate output tokens while largely preserving task utility. We define such manipulation as a Provider-Side Token Inflation Attack (PTIA) and instantiate five representative attacks at the query, prompt, representation, and model levels of the provider-controlled pipeline. Our experiments show that each attack increases mean output length to more than 10.2x the clean baseline, demonstrating PTIA's financial appeal and feasibility at multiple stages of generation. Yet auditing PTIA from black-box responses is difficult for users. Our key observation is PTIA saturation: an initial attack sharply lengthens output, but further strengthening or composition has much less effect. We trace this saturation to stopping behavior: an initial PTIA sharply lowers the end-of-sequence token probability, whereas further intervention lowers it only marginally. Building on this insight, we design a lightweight single-probe audit that applies a controlled lengthening intervention. Under PTIA, the probe induces far fewer additional tokens than under normal service. The audit requires neither a trusted local reference model nor historical clean responses, and its separately issued original and probed requests resemble ordinary traffic, making evasion difficult. Across four open-weight models, it achieves an average detection rate of 85.1% with false-positive rates below 2%. Across 15 real LLM API services, the audit flags 7 for PTIA-consistent behavior.

Thu 17 SeptCryptography and Security
The gist
In some language model services, users pay based on how many words the model generates. The authors found that some providers might secretly increase the length of responses to make users pay more, without clearly changing helpfulness. They designed several ways this can be done and showed it’s practical. They also created a simple test that can detect when such manipulations happen, even without knowing the original clean response. When tested on multiple models and real services, this test flagged several services likely inflating output length.
Open → 2609.20370v1