Danilo Francati

Hanlin Zhang, Benjamin L. Edelman, Danilo Francati et al.

Watermarking generative models consists of planting a statistical signal (watermark) in a model's output so that it can be later verified that the output was generated by the given model. A strong watermarking scheme satisfies the property that a computationally bounded attacker cannot erase the watermark without causing significant quality degradation. In this paper, we study the (im)possibility of strong watermarking schemes. We prove that, under well-specified and natural assumptions, strong watermarking is impossible to achieve. This holds even in the private detection algorithm setting, where the watermark insertion and detection algorithms share a secret key, unknown to the attacker. To prove this result, we introduce a generic efficient watermark attack; the attacker is not required to know the private key of the scheme or even which scheme is used. Our attack is based on two assumptions: (1) The attacker has access to a "quality oracle" that can evaluate whether a candidate output is a high-quality response to a prompt, and (2) The attacker has access to a "perturbation oracle" which can modify an output with a nontrivial probability of maintaining quality, and which induces an efficiently mixing random walk on high-quality outputs. We argue that both assumptions can be satisfied in practice by an attacker with weaker computational capabilities than the watermarked model itself, to which the attacker has only black-box access. Furthermore, our assumptions will likely only be easier to satisfy over time as models grow in capabilities and modalities. We demonstrate the feasibility of our attack by instantiating it to attack three existing watermarking schemes for large language models: Kirchenbauer et al. (2023), Kuditipudi et al. (2023), and Zhao et al. (2023). The same attack successfully removes the watermarks planted by all three schemes, with only minor quality degradation.

Danilo Francati, Yevin Nikhel Goonatilake, Shubham Pawar et al.

We prove a sharp threshold for the robustness of cryptographic watermarking for generative models. This is achieved by introducing a coding abstraction, which we call messageless secret-key codes, that formalizes sufficient and necessary requirements of robust watermarking: soundness, tamper detection, and pseudorandomness. Thus, we establish that robustness has a precise limit: For binary outputs no scheme can survive if more than half of the encoded bits are modified, and for an alphabet of size q the corresponding threshold is $(1-1/q)$ of the symbols. Complementing this impossibility, we give explicit constructions that meet the bound up to a constant slack. For every $δ > 0$, assuming pseudorandom functions and access to a public counter, we build linear-time codes that tolerate up to $(1/2)(1-δ)$ errors in the binary case and $(1-1/q)(1-δ)$ errors in the $q$-ary case. Together with the lower bound, these yield the maximum robustness achievable under standard cryptographic assumptions. We then test experimentally whether this limit appears in practice by looking at the recent watermarking for images of Gunn, Zhao, and Song (ICLR 2025). We show that a simple crop and resize operation reliably flipped about half of the latent signs and consistently prevented belief-propagation decoding from recovering the codeword, erasing the watermark while leaving the image visually intact. These results provide a complete characterization of robust watermarking, identifying the threshold at which robustness fails, constructions that achieve it, and an experimental confirmation that the threshold is already reached in practice.

Dario Pasquini, Danilo Francati, Giuseppe Ateniese

Secure aggregation is a cryptographic protocol that securely computes the aggregation of its inputs. It is pivotal in keeping model updates private in federated learning. Indeed, the use of secure aggregation prevents the server from learning the value and the source of the individual model updates provided by the users, hampering inference and data attribution attacks. In this work, we show that a malicious server can easily elude secure aggregation as if the latter were not in place. We devise two different attacks capable of inferring information on individual private training datasets, independently of the number of users participating in the secure aggregation. This makes them concrete threats in large-scale, real-world federated learning applications. The attacks are generic and equally effective regardless of the secure aggregation protocol used. They exploit a vulnerability of the federated learning protocol caused by incorrect usage of secure aggregation and lack of parameter validation. Our work demonstrates that current implementations of federated learning with secure aggregation offer only a "false sense of security".

Danilo Francati, Giuseppe Ateniese, Abdoulaye Faye et al.

The cloud changed the way we manage and store data. Today, cloud storage services offer clients an infrastructure that allows them a convenient source to store, replicate, and secure data online. However, with these new capabilities also come limitations, such as lack of transparency, limited decentralization, and challenges with privacy and security. And, as the need for more agile, private and secure data solutions continues to grow exponentially, rethinking the current structure of cloud storage is mission-critical for enterprises. By leveraging and building upon blockchain's unique attributes, including immutability, security to the data element level, distributed (no single point of failure), we have developed a solution prototype that allows data to be reliably stored while simultaneously being secured, with tamper-evident auditability, via blockchain. The result, Audita, is a flexible solution that assures data protection and solves challenges such as scalability and privacy. Audita works via an augmented blockchain network of participants that include storage-nodes and block-creators. In addition, it provides an automatic and fair challenge system to assure that data is distributed and reliably and provably stored. While the prototype is built on Quorum, the solution framework can be used with any blockchain platform. The benefit is a system that is built to grow along with the data needs of enterprises, while continuing to build the network via incentives and solving for issues such as auditing and outsourcing.

Adriano Di Luzio, Danilo Francati, Giuseppe Ateniese

This work presents Arcula, a new design for hierarchical deterministic wallets that brings identity-based addresses to the blockchain. Arcula is built on top of provably secure cryptographic primitives. It generates all its cryptographic secrets from a user-provided seed and enables the derivation of new public keys based on the identities of users, without requiring any secret information. Unlike other wallets, it achieves all these properties while being secure against privilege escalation. We formalize the security model of hierarchical deterministic wallets and prove that an attacker compromising an arbitrary number of users within an Arcula wallet cannot escalate his privileges and compromise users higher in the access hierarchy. Our design works out-of-the-box with any blockchain that enables the verification of signatures on arbitrary messages. We evaluate its usage in a real-world scenario on the Bitcoin Cash network.