arXiv Machine Learning

Single-Query Black-Box Calibration Auditing via Logit Bias

The paper introduces a method for auditing the calibration of large language models (LLMs) that only exposes a logit_bias parameter. By mathematically manipulating this parameter, the authors can evaluate exact probability thresholds with a single query per sample, enabling a provably consistent estimator of True Calibration Error for binary tasks. This approach offers an efficient framework for auditing black‑box foundation models despite limited access to continuous output probabilities.

arXiv AI
2d ago

Pinocchio: Fast Uncertainty Estimates for Black-Box Language Models

Pinocchio is an external calibrator that provides fast uncertainty estimates for black-box language models. It predicts the correctness of responses from seven trained LLMs with an AUROC of 0.862 and can transfer zero‑shot to thirteen unseen models from eight organizations. The method requires only a single forward pass and no access to the target model’s internal states, and a lightweight 0.8B checkpoint achieves comparable performance.

By Kevin David Hayes, Arka Pal, Haosong Zhang, Tom Goldstein, Micah Goldblum
arXiv Machine Learning
Sep 11

Black-Box Membership Inference via Word-Level Probability Estimation

The paper introduces Word-level Probability MIA (WPMIA), a black-box membership inference attack that estimates word-level generation probabilities via Monte Carlo sampling and local kernel smoothing, then aggregates them into a sequence-level likelihood estimator. By conditioning on different prefixes, WPMIA amplifies distributional differences between member and non-member texts, outperforming existing black-box baselines on open-source LLMs and achieving an average TPR@5%FPR of 42.0 on proprietary models such as GPT‑5‑Chat, Gemini‑2.5‑Flash, and Claude‑4.5‑Haiku.

By Shengjie Niu, Yeheng Ge, Jian Huang
arXiv AI
Aug 19

Too Sure to Be Safe: Model Calibration for Reliable Log Anomaly Detection

The paper introduces LoRD, a lightweight post‑hoc calibration framework designed to improve confidence reliability in language‑model‑based log anomaly detectors. LoRD learns route‑specific reliability models from latent representations of correctly classified validation samples and uses reconstruction distances to estimate prediction reliability. By selectively recalibrating high‑risk predictions, LoRD reduces overconfident errors while maintaining strong anomaly detection performance across four large‑scale log benchmark datasets.

By Bin Li, Dongdong Wang, Siyang Lu
arXiv Computation and Language
Aug 28

An Embarrassingly Simple Detector for Model Extraction Attacks in Large Language Model API Traffic

The paper presents a simple detector for model extraction attacks on large language model APIs. It frames detection as a benign‑calibrated traffic‑window distribution test, embedding queries into a semantic space and using maximum mean discrepancy (MMD) to compare against historical benign traffic. Evaluated on fourteen attacker‑normal query pairs across four extraction scenarios, MMD achieves near‑perfect true‑positive rates while maintaining a 0.3% false‑positive rate, outperforming several existing baselines.

By Shuze Liu, Qianwen Guo, Yushun Dong
arXiv Computation and Language
2d ago

Calibration as a First-Class Criterion in LLM Evaluation

The paper argues that calibration—how well a language model’s confidence aligns with its actual correctness—should be a standard evaluation metric for large language models (LLMs). It notes that while calibration metrics exist, they are rarely applied outside specialized NLP subfields, leading to unverified confidence scores in new models, datasets, and benchmarks. The authors highlight the risks of miscalibration both at deployment (overconfident errors causing harm) and in research workflows (affecting LLM-as-a-judge, synthetic data generation, and active learning). They call for every NLP subfield to pair its primary performance metric with a calibration score, treating calibration as an essential property of every model.

By Mario Sanz-Guerrero, Katharina von der Wense