Roger Wattenhofer

Mose Mizrahi, Roger Wattenhofer

We consider an asynchronous network of $n$ parties connected to each other via secure channels, up to $t$ of which are byzantine. We study common coin tossing, a task where the parties try to agree on an unpredictable random value, with some chance of failure due to the byzantine parties' influence. Coin tossing is a well-known and often-studied task due to its use in byzantine agreement. In this work, we present a committee-based method to transform strong (rarely failing) binary common coins into weaker ones that asymptotically require less communication. For any $k > 2$ and $\varepsilon > 0$, we can transform a strong binary coin that costs $\widetilde{O}(n^k)$ bits of communication into a weak binary coin that costs $\widetilde{O}(\varepsilon^{-2k}n^{3 - 2/k})$ bits. This latter coin tolerates $\varepsilon n$ fewer byzantine parties than the strong coin it is based on, and it fails with an arbitrarily small constant probability. With our method, we obtain a secure-channel-based perfectly secure coin for $t \leq (\frac{1}{4} - \varepsilon)n$ faults that costs $\widetilde{O}(n^{2.5})$ bits, as well as a coin based on cryptographic hashing for $t \leq (\frac{1}{3} - \varepsilon)n$ faults that costs $\widetilde{O}(n^{7/3}κ)$ bits. These are to our knowledge the first PKI-free asynchronous common coins that cost $o(n^3)$ bits of communication but still succeed with at least constant probability against $t = Θ(n)$ adaptive byzantine faults.

Frédéric Berdoz, Leonardo Rugli, Roger Wattenhofer

Large language models are increasingly deployed as cooperating agents, yet their behavior in adversarial consensus settings has not been systematically studied. We evaluate LLM-based agents on a Byzantine consensus game over scalar values using a synchronous all-to-all simulation. We test consensus in a no-stake setting where agents have no preferences over the final value, so evaluation focuses on agreement rather than value optimality. Across hundreds of simulations spanning model sizes, group sizes, and Byzantine fractions, we find that valid agreement is not reliable even in benign settings and degrades as group size grows. Introducing a small number of Byzantine agents further reduces success. Failures are dominated by loss of liveness, such as timeouts and stalled convergence, rather than subtle value corruption. Overall, the results suggest that reliable agreement is not yet a dependable emergent capability of current LLM-agent groups even in no-stake settings, raising caution for deployments that rely on robust coordination.

Frédéric Berdoz, Yann Billeter, Yann Vonlanthen et al.

Opinion modeling aims to capture individual or group political preferences, enabling applications such as digital democracies, where models could help shape fairer and more popular policies. Given their versatility, strong generalization capabilities, and demonstrated success across diverse text-to-text applications, large language models (LLMs) are natural candidates for this task. However, due to their statistical nature and limited causal understanding, they tend to produce biased opinions when prompted naively. In this work, we study whether reasoning can improve opinion alignment. Motivated by the recent advancement in mathematical reasoning enabled by reinforcement learning (RL), we train models to produce profile-consistent answers through structured reasoning. We evaluate our approach on three datasets covering U.S., European, and Swiss politics. Results indicate that reasoning enhances opinion modeling and is competitive with strong baselines, but does not fully remove bias, highlighting the need for additional mechanisms to build faithful political digital twins using LLMs. By releasing both our method and datasets, we establish a solid baseline to support future research on LLM opinion alignment.

Isotta Magistrali, Frédéric Berdoz, Sam Dauncey et al.

As AI systems approach superhuman capabilities, scalable oversight increasingly relies on LLM-as-a-judge frameworks where models evaluate and guide each other's training. A core assumption is that binary preference labels provide only semantic supervision about response quality. We challenge this assumption by demonstrating that preference labels can function as a covert communication channel. We show that even when a neutral student model generates semantically unbiased completions, a biased judge can transmit unintended behavioral traits through preference assignments, which even strengthen across iterative alignment rounds. Our findings suggest that robust oversight in superalignment settings requires mechanisms that can detect and mitigate subliminal preference transmission, particularly when judges may pursue unintended objectives.

Frédéric Berdoz, Luca A. Lanzendörfer, Kaan Bayraktar et al.

Large Language Models (LLMs) are increasingly deployed across diverse applications, raising critical questions for governance, accountability, and data provenance. Understanding which training data most influenced a model's output remains a fundamental open problem. We address this challenge through training data attribution (TDA) for auto-regressive LLMs by expanding upon the inverse formulation: How would training data be affected if the model had seen the generated output during training? Our method perturbs the base model using bidirectional gradient optimization (gradient ascent and descent) on a generated text sample and measures the resulting change in loss across training samples. Our framework supports attribution at arbitrary data granularity, enabling both factual and stylistic attribution. We evaluate our method against baselines on pretrained models with known datasets, and show that it outperforms previous work on influence metrics, thereby enhancing model interpretability, an essential requirement for accountable AI systems.