Whether language models can systematically generalize remains actively debated. Yet empirical performance is jointly shaped by multiple factors such as training data, training paradigms, and inference-time strategies, making failures difficult to interpret. We introduce a controlled synthetic environment based on shortest-path planning, a canonical composable sequential optimization problem. The setup enables clean separation of these factors and supports two orthogonal axes of generalization: spatial transfer to unseen maps and length scaling to longer-horizon problems. We find that models exhibit strong spatial transfer but consistently fail under length scaling due to recursive instability. We further analyze how distinct stages of the learning pipeline influence systematic problem-solving: for example, data coverage sets capability limits; reinforcement learning improves training stability but does not expand those limits; and inference-time scaling enhances performance but cannot rescue length-scaling failures.

Someone who learns to walk shortest paths in New York can, upon receiving a map of Paris, immediately apply the same rule to navigate, despite never practicing there. This ability to recombine known rules to solve novel problems exemplifies compositional generalization (CG), a hallmark of human cognition. Yet our understanding of what drives the success or failure of such extrapolative problem solving, particularly the roles of training data properties and optimization paradigms, remains limited. In this work, we introduce a controlled map-navigation testbed that cleanly separates two dimensions of CG: ***spatial transfer*** (systematicity across environments) and ***length scaling*** (productivity along problem difficulty). Through quantitative experiments, we show that transfer is enabled by sufficient distinct questions with high coverage and modest diversity, while scaling critically depends on exposure to neighboring-but-longer examples. Finally, we find that reinforcement learning (RL) stabilizes optimization but does not surpass the ceiling set by supervised fine-tuning (SFT). Together, these results provide principled insights into how data properties and training paradigms shape extrapolative problem solving.

Distillation attacks create a deployment trade-off for model providers: the same outputs that make a model more useful can also make it easier to imitate. We study this trade-off through a minimax game between a utility-constrained teacher and an adaptive student. Our framework yields tractable one-sided response rules: an adaptive evaluation rule in which the student reweights high-value examples, and a teacher-side defense template that suppresses outputs most useful for distillation. From a cheap proxy for example value, we derive Product-of-Experts (PoE), a simple forward-pass-only defense that combines the teacher with a proxy student during generation. Empirically, adaptive evaluation reveals a large passive–adaptive gap: on state-of-the-art defenses, adaptive students recover substantially more capability than passive evaluation suggests on GSM8K and MATH. Under this stronger evaluation, the apparent robustness gap between expensive defenses and PoE narrows considerably, while PoE remains substantially cheaper and preserves higher-quality reasoning traces. Overall, our results suggest that strong distillation remains difficult to stop, and that progress on antidistillation should be judged against adaptive students rather than passive ones. Our code is available at: https://github.com/ysfalh/distillation-game.

AbstractIncentivization in clinical trial participation can be challenging, with many studies failing to meet recruitment or retention goals despite traditional compensation strategies. Digital health evolves, and with it, new approaches can emerge to engage participants meaningfully. We propose unique digital images as a novel, symbolic incentive for clinical trials. Digital images combine qualities such as personalization, ownership, and digital visibility, which may drive engagement more effectively than monetary rewards alone. In our illustrative study, participants complete artificial intelligence–personalized digital therapeutics training using CURATE.DTx, generating individualized learning trajectories. These are transformed into digital artworks and minted as nonfungible tokens given as a reward upon trial completion. This concept integrates gamification, personalization, and blockchain technology to support both intrinsic and extrinsic motivation. We explore the implications for decentralized health care, long-term behavior change, and participant recognition in the context of preventive medicine and longevity science. Our aim is to encourage research into the use of digital incentives to transform the trial participant experience and promote sustained engagement in health interventions.

Machine learning models are known to leak sensitive information, as they inevitably memorize (parts of) their training data. More alarmingly, large language models (LLMs) are now trained on nearly all available data, which amplifies the magnitude of information leakage and raises serious privacy risks. Hence, it is more crucial than ever to quantify privacy risk before the release of LLMs. The standard method to quantify privacy is via membership inference attacks, where the state-of-the-art approach is the Robust Membership Inference Attack (RMIA). In this paper, we present InfoRMIA, a principled information-theoretic formulation of membership inference. Our method consistently outperforms RMIA across benchmarks while also offering improved computational efficiency. In the second part of the paper, we identify the limitations of treating sequence-level membership inference as the gold standard for measuring leakage. We propose a new perspective for studying membership and memorization in LLMs: token-level signals and analyses. We show that a simple token-based InfoRMIA can pinpoint which tokens are memorized within generated outputs, thereby localizing leakage from the sequence level down to individual tokens, while achieving stronger sequence-level inference power on LLMs. This new scope rethinks privacy in LLMs and can lead to more targeted mitigation, such as exact unlearning.

Large language models (LLMs) are known to "hallucinate" by generating false or misleading outputs. Hallucinations pose various harms, from erosion of trust to widespread misinformation. Existing hallucination evaluation, however, focuses only on correctness and often overlooks consistency, necessary to distinguish and address these harms. To bridge this gap, we introduce prompt multiplicity, a framework for quantifying consistency in LLM evaluations. Our analysis reveals significant multiplicity (over 50

Membership Inference Attacks (MIAs) on pre-trained Large Language Models (LLMs) aim at determining if a data point was part of the model's training set. Prior MIAs that are built for classification models fail at LLMs, due to ignoring the generative nature of LLMs across token sequences. In this paper, we present a novel attack on pre-trained LLMs that adapts MIA statistical tests to the perplexity dynamics of subsequences within a data point. Our method significantly outperforms prior approaches, revealing context-dependent memorization patterns in pre-trained LLMs.

Machine learning models are known to leak sensitive information, as they inevitably memorize (parts of) their training data. More alarmingly, large language models (LLMs) are now trained on nearly all available data, which amplifies the magnitude of information leakage and raises serious privacy risks. Hence, it is more crucial than ever to quantify privacy risk before the release of LLMs. The standard method to quantify privacy is via membership inference attacks, where the state-of-the-art approach is the Robust Membership Inference Attack (RMIA). In this paper, we present InfoRMIA, a principled information-theoretic formulation of membership inference. Our method consistently outperforms RMIA across benchmarks while also offering improved computational efficiency. In the second part of the paper, we identify the limitations of treating sequence-level membership inference as the gold standard for measuring leakage. We propose a new perspective for studying membership and memorization in LLMs: token-level signals and analyses. We show that a simple token-based InfoRMIA can pinpoint which tokens are memorized within generated outputs, thereby localizing leakage from the sequence level down to individual tokens, while achieving stronger sequence-level inference power on LLMs. This new scope rethinks privacy in LLMs and can lead to more targeted mitigation, such as exact unlearning.

As the era of autonomous agents making decisions on behalf of users unfolds, ensuring contextual integrity (CI) -- what is the appropriate information to share while carrying out a certain task -- becomes a central question to the field. We posit that CI demands a form of reasoning where the agent needs to reason about the context in which it is operating. To test this, we first prompt LLMs to reason explicitly about CI when deciding what information to disclose. We then extend this approach by developing a reinforcement learning (RL) framework that further instills in models the reasoning necessary to achieve CI. Using a synthetic, automatically created, dataset of only $\sim700$ examples but with diverse contexts and information disclosure norms, we show that our method substantially reduces inappropriate information disclosure while maintaining task performance across multiple model sizes and families. Importantly, improvements transfer from this synthetic dataset to established CI benchmarks such as PrivacyLens that has human annotations and evaluates privacy leakage of AI assistants in actions and tool calls.

The principle of data minimization aims to reduce the amount of data collected, processed or retained to minimize the potential for misuse, unauthorized access, or data breaches. Rooted in privacy-by-design principles, data minimization has been endorsed by various global data protection regulations. However, its practical implementation remains a challenge due to the lack of a rigorous formulation. This paper addresses this gap and introduces an optimization framework for data minimization based on its legal definitions. It then adapts several optimization algorithms to perform data minimization and conducts a comprehensive evaluation in terms of their compliance with minimization objectives as well as their impact on user privacy. Our analysis underscores the mismatch between the privacy expectations of data minimization and the actual privacy benefits, emphasizing the need for approaches that account for multiple facets of real-world privacy risks.

How much of my data was used to train a machine learning model? This is a critical question for data owners assessing the risk of unauthorized usage of their data to train models. However, previous work mistakenly treats this as a binary problem—inferring whether all-or-none or any-or-none of the data was used—which is fragile when faced with real, non-binary data usage risks. To address this, we propose a fine-grained analysis called Dataset Usage Cardinality Inference (DUCI), which estimates the exact proportion of data used. Our algorithm, leveraging debiased membership guesses, matches the performance of the optimal MLE approach (with a maximum error <0.1) but with significantly lower (e.g., $300 \times$ less) computational cost.

Machine learning models can leak private information about their training data. The standard methods to measure this privacy risk, based on membership inference attacks (MIAs), only check if a given data point exactly matches a training point, neglecting the potential of similar or partially overlapping memorized data revealing the same private information. To address this issue, we introduce the class of range membership inference attacks (RaMIAs), testing if the model was trained on any data in a specified range (defined based on the semantics of privacy). We formulate the RaMIAs game and design a principled statistical test for its composite hypotheses. We show that RaMIAs can capture privacy loss more accurately and comprehensively than MIAs on various types of data, such as tabular, image, and language. RaMIA paves the way for more comprehensive and meaningful privacy auditing of machine learning algorithms.

How much information about training samples can be leaked through synthetic data generated by Large Language Models (LLMs)? Overlooking the subtleties of information flow in synthetic data generation pipelines can lead to a false sense of privacy. In this paper, we assume an adversary has access to some synthetic data generated by a LLM. We design membership inference attacks (MIAs) that target the training data used to fine-tune the LLM that is then used to synthesize data. The significant performance of our MIA shows that synthetic data leak information about the training data. Further, we find that canaries crafted for model-based MIAs are sub-optimal for privacy auditing when only synthetic data is released. Such out-of-distribution canaries have limited influence on the model’s output when prompted to generate useful, in-distribution synthetic data, which drastically reduces their effectiveness. To tackle this problem, we leverage the mechanics of auto-regressive models to design canaries with an in-distribution prefix and a high-perplexity suffix that leave detectable traces in synthetic data. This enhances the power of data-based MIAs and provides a better assessment of the privacy risks of releasing synthetic data generated by LLMs.

How effectively can LLM-based AI assistants utilize their memory (context) to perform various tasks? Traditional data benchmarks, which are often manually crafted, suffer from several limitations: they are static, susceptible to overfitting, difficult to interpret, and lack actionable insights--failing to pinpoint the specific capabilities a model lacks when it does not pass a test. In this paper, we present a framework for automatically generating a comprehensive set of tests to evaluate models' abilities to use their memory effectively. Our framework extends the range of capability tests beyond the commonly explored (passkey, key-value, needle in the haystack) search, a dominant focus in the literature. Specifically, we evaluate models on atomic tasks such as searching, recalling, editing, matching, comparing information in context memory, performing basic operations when inputs are structured into distinct blocks, and maintaining state while operating on memory, simulating real-world data. Additionally, we design composite tests to investigate the models' ability to perform more complex, integrated tasks. Our benchmark enables an interpretable, detailed assessment of memory capabilities of LLMs.

Watermarking is a key technique for detecting AI-generated text. In this work, we study its vulnerabilities and introduce the Smoothing Attack, a novel watermark removal method. By leveraging the relationship between the model’s confidence and watermark detectability, our attack selectively smoothes the watermarked content, erasing watermark traces while preserving text quality. We validate our attack on open-source models ranging from 1.3 B to 30B parameters on 10 different watermarks, demonstrating its effectiveness. Our findings expose critical weaknesses in existing watermarking schemes and highlight the need for stronger defenses.

We introduce an analytical framework to quantify the changes in a machine learning algorithm's output distribution following the inclusion of a few data points in its training set, a notion we define as leave-one-out distinguishability (LOOD). This is key to measuring data **memorization** and information **leakage** as well as the **influence** of training data points in machine learning. We illustrate how our method broadens and refines existing empirical measures of memorization and privacy risks associated with training data. We use Gaussian processes to model the randomness of machine learning algorithms, and validate LOOD with extensive empirical analysis of leakage using membership inference attacks. Our analytical framework enables us to investigate the causes of leakage and where the leakage is high. For example, we analyze the influence of activation functions, on data memorization. Additionally, our method allows us to identify queries that disclose the most information about the training data in the leave-one-out setting. We illustrate how optimal queries can be used for accurate **reconstruction** of training data.

Neural recording systems play a crucial role in comprehending the intricacies of the brain and advancing treatments for neurological disorders. Within these systems, the analog-to-digital converter (ADC) serves as a fundamental component, converting the electrical signals from the brain into digital data that can be further processed and analyzed by computing units. This research introduces a novel nonlinear ADC designed specifically for spike sorting in biomedical applications. Employing MOSFET varactors and voltage-controlled oscillators (VCOs), this ADC exploits the nonlinear capacitance properties of MOSFET varactors, achieving a parabolic quantization function that digitizes the noise with low resolution and the spikes with high resolution, effectively suppressing the background noise present in biomedical signals. This research aims to develop a reconfigurable, nonlinear voltage-controlled oscillator (VCO)-based ADC, specifically designed for implantable neural recording systems used in neuroprosthetics and brain–machine interfaces. The proposed design enhances the signal-to-noise ratio and reduces power consumption, making it more efficient for real-time neural data processing. By improving the performance and energy efficiency of these devices, the research contributes to the development of more reliable medical technologies for monitoring and treating neurological disorders. The quantization step of the ADC spans from 44.8 mV in the low-amplitude range to 1.4 mV in the high-amplitude range. The circuit was designed and simulated utilizing a 180 nm CMOS process; however, no physical prototype has been fabricated at this stage. Post-layout simulations confirm the expected performance. Occupying a silicon area is 0.09 mm2. Operating at a sampling frequency of 16 kS/s and a supply voltage of 1 volt, this ADC consumes 62.4 µW.

The interactive use of large language models (LLMs) in AI assistants (at work, home, etc.) introduces a new set of inference-time privacy risks: LLMs are fed different types of information from multiple sources in their inputs and are expected to reason about what to share in their outputs, for what purpose and with whom, within a given context. In this work, we draw attention to the highly critical yet overlooked notion of contextual privacy by proposing ConfAIde, a benchmark designed to identify critical weaknesses in the privacy reasoning capabilities of instruction-tuned LLMs. Our experiments show that even the most capable models such as GPT-4 and ChatGPT reveal private information in contexts that humans would not, 39% and 57% of the time, respectively. This leakage persists even when we employ privacy-inducing prompts or chain-of-thought reasoning. Our work underscores the immediate need to explore novel inference-time privacy-preserving approaches, based on reasoning and theory of mind.

Membership inference attacks aim to detect if a particular data point was used in training a model. We design a novel statistical test to perform robust membership inference attacks (RMIA) with low computational overhead. We achieve this by a fine-grained modeling of the null hypothesis in our likelihood ratio tests, and effectively leveraging both reference models and reference population data samples. RMIA has superior test power compared with prior methods, throughout the TPR-FPR curve (even at extremely low FPR, as low as 0). Under computational constraints, where only a limited number of pre-trained reference models (as few as 1) are available, and also when we vary other elements of the attack (e.g., data distribution), our method performs exceptionally well, unlike prior attacks that approach random guessing. RMIA lays the groundwork for practical yet accurate data privacy risk assessment in machine learning.