Papers for

data compliance officers

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.

Tabular foundation models can reveal if a dataset trained them

Dataset Inference for Data Provenance and Privacy Auditing in Tabular Foundation Models

Abstract: Tabular foundation models (TFMs) are increasingly deployed through black-box APIs and trained on large collections of real-world tabular datasets. As this training paradigm becomes more widely adopted, private or proprietary datasets may be incorporated into pre-training corpora, yet there are currently no dedicated methods for determining whether a given tabular dataset was used to train a TFM. As a solution, we introduce the first dataset inference method for TFMs, aiming to infer whether a suspect dataset was part of a model's pre-training data. We systematically analyze a broad collection of candidate signals that can be observed from a black-box TFM via input manipulation and find that individual signals provide only weak evidence of training data membership. However, by combining multiple complementary signals, we can infer dataset membership for several state-of-the-art TFMs trained on real tabular data, achieving up to 0.99 ROCAUC dependent on the TFM. We then study factors that influence dataset identification, including pre-training data composition, model capacity, and the use of real vs. synthetic data. Our results show that our dataset inference method is a practical auditing tool for detecting privacy leakage and the use of proprietary datasets in TFMs.

Fri 16 OctMachine Learning
The gist
Tabular foundation models are trained on many datasets, sometimes including private data, but it's hard to know which datasets were used. The authors created a method to check if a particular dataset was included in the training of such a model by probing the model and combining different clues. This method can identify dataset usage with high accuracy for various models on real-world tabular data. Their study also shows what factors affect this ability, helping with privacy checks and verifying proprietary data use.
Open → 2610.24278v1

Surprise masking reduces private data leaks in language models

Surprise Masking: Private Data Filtering with Model Likelihoods

Abstract: Data useful for language model (LM) training can contain private or unwanted information that is at risk of being learned or memorized. We propose Surprise Masking (SM), a simple form of perplexity filtering that uses a publicly available reference model to identify unwanted spans and suppress learning from them while preserving the rest of each example. On synthetic code benchmarks, we show SM improves the privacy-utility tradeoff over standard masking methods and PII filtering. In fine-tuning experiments, SM reduces approximate extraction of private values from revealed prefixes from 66.3% to 2.3% and the reproduction of stylistic coding quirks from 23.6% to 1.3%, both while preserving performance on the evaluated coding tasks. We also evaluate SM for sanitizing datasets by directly removing unwanted content. Here, SM outperforms DePA, a perplexity-based defense for detecting poisoned dead code, while specialized detectors perform better on broader credential and PII benchmarks. Across eleven settings, we find that SM's downstream success is predictable from the separation of unwanted and useful content in the distribution, and it is particularly useful with code data. Ultimately, SM is an inexpensive, adaptable method to preserve valuable examples while reducing the learning of unwanted information.

Fri 16 OctCryptography and Security
The gist
Training language models can accidentally make them memorize private or unwanted information from the data. The authors propose Surprise Masking, a way to use a public model to spot and mask these sensitive parts so the model learns from the rest safely. Their tests show it lowers privacy risks far more than standard methods while keeping good performance on coding tasks. It works best when unwanted parts can be clearly separated from useful data and is especially helpful for code datasets.
Open → 2610.22988v1