Showing papers presented at "USENIX Security Symposium in 2023"

LZR: Identifying Unexpected Internet Services

Abstract: Internet-wide scanning is a commonly used research technique that has helped uncover real-world attacks, find cryptographic weaknesses, and understand both operator and mis-creant behavior. Studies that employ scanning have largely assumed that services are hosted on their IANA-assigned ports, overlooking the study of services on unusual ports. In this work, we investigate where Internet services are deployed in practice and evaluate the security posture of services on unexpected ports. We show protocol deployment is more diffuse than previously believed and that protocols run on many additional ports beyond their primary IANA-assigned port. For example, only 3% of HTTP and 6% of TLS services run on ports 80 and 443, respectively. Services on non-standard ports are more likely to be insecure, which results in studies dramatically underestimating the security posture of Internet hosts. Building on our observations, we introduce LZR (“Laser”), a system that identifies 99% of identifiable unexpected services in five handshakes and dramatically reduces the time needed to perform application-layer scans on ports with few responsive expected services (e.g., 5500% speedup on 27017/MongoDB). We conclude with recommendations for future studies.

Token-Level Fuzzing

Abstract: Fuzzing has become a commonly used approach to identifying bugs in complex, real-world programs. However, interpreters are notoriously difficult to fuzz effectively, as they expect highly structured inputs, which are rarely produced by most fuzzing mutations. For this class of programs, grammar-based fuzzing has been shown to be effective. Tools based on this approach can find bugs in the code that is executed after parsing the interpreter inputs, by following language-specific rules when generating and mutating test cases. Unfortunately, grammar-based fuzzing is often unable to discover subtle bugs associated with the parsing and handling of the language syntax. Additionally, if the grammar provided to the fuzzer is incomplete, or does not match the implementation completely, the fuzzer will fail to exercise important parts of the available functionality. In this paper, we propose a new fuzzing technique, called Token-Level Fuzzing. Instead of applying mutations either at the byte level or at the grammar level, Token-Level Fuzzing applies mutations at the token level. Evolutionary fuzzers can leverage this technique to both generate inputs that are parsed successfully and generate inputs that do not conform strictly to the grammar. As a result, the proposed approach can find bugs that neither byte-level fuzzing nor grammar-based fuzzing can find. We evaluated Token-Level Fuzzing by modifying AFL and fuzzing four popular JavaScript engines, finding 29 previously unknown bugs, several of which could not be found with state-of-the-art byte-level and grammar-based fuzzers.

A Highly Accurate Query-Recovery Attack against Searchable Encryption using Non-Indexed Documents

Abstract: Cloud data storage solutions offer customers cost-effective and reduced data management. While attractive, data security issues remain to be a core concern. Traditional encryption protects stored documents, but hinders simple functionalities such as keyword search. Therefore, searchable encryption schemes have been proposed to allow for the search on encrypted data. Efficient schemes leak at least the access pattern (the accessed documents per keyword search), which is known to be exploitable in query recovery attacks assuming the attacker has a significant amount of background knowledge on the stored documents. Existing attacks can only achieve decent results with strong adversary models (e.g. at least 20% of previously known documents or require additional knowledge such as on query frequencies) and they give no metric to evaluate the certainty of recovered queries. This hampers their practical utility and questions their relevance in the real-world. We propose a refined score attack which achieves query recovery rates of around 85% without requiring exact background knowledge on stored documents; a distributionally similar, but otherwise different (i.e., non-indexed), dataset suffices. The attack starts with very few known queries (around 10 known queries in our experiments over different datasets of varying size) and then iteratively recovers further queries with confidence scores by adding previously recovered queries that had high confidence scores to the set of known queries. Additional to high recovery rates, our approach yields interpretable results in terms of confidence scores.

Pool Inference Attacks on Local Differential Privacy: Quantifying the Privacy Guarantees of Apple's Count Mean Sketch in Practice

Abstract: Behavioral data generated by users' devices, ranging from emoji use to pages visited, are collected at scale to improve apps and services. These data, however, contain fine-grained records and can reveal sensitive information about individual users. Local differential privacy has been used by companies as a solution to collect data from users while preserving privacy. We here first introduce pool inference attacks, where an adversary has access to a user's obfuscated data, defines pools of objects, and exploits the user's polarized behavior in multiple data collections to infer the user's preferred pool. Second, we instantiate this attack against Count Mean Sketch, a local differential privacy mechanism proposed by Apple and deployed in iOS and Mac OS devices, using a Bayesian model. Using Apple's parameters for the privacy loss $\varepsilon$, we then consider two specific attacks: one in the emojis setting -- where an adversary aims at inferring a user's preferred skin tone for emojis -- and one against visited websites -- where an adversary wants to learn the political orientation of a user from the news websites they visit. In both cases, we show the attack to be much more effective than a random guess when the adversary collects enough data. We find that users with high polarization and relevant interest are significantly more vulnerable, and we show that our attack is well-calibrated, allowing the adversary to target such vulnerable users. We finally validate our results for the emojis setting using user data from Twitter. Taken together, our results show that pool inference attacks are a concern for data protected by local differential privacy mechanisms with a large $\varepsilon$, emphasizing the need for additional technical safeguards and the need for more research on how to apply local differential privacy for multiple collections.