Aurora: Are Android Malware Classifiers Reliable under Distribution Shift?

The performance figures of modern drift-adaptive malware classifiers appear promising, but does this translate to genuine operational reliability? The standard evaluation paradigm primarily focuses on baseline performance metrics, neglecting confidence-error alignment and operational stability...

Authorization Bypass

github.com/openfga/openfga is vulnerable to authorization bypass. The vulnerability is due to incorrect permission evaluation when usersets and type-bound public access overlap without proper tuple assignments, allowing attackers unauthorized resource access or actions...

System Prompt Extraction Attacks and Defenses in Large Language Models

The system prompt in Large Language Models LLMs plays a pivotal role in guiding model behavior and response generation. Often containing private configuration details, user roles, and operational instructions, the system prompt has become an emerging attack target. Recent studies have shown that...

The Feasibility of Topic-Based Watermarking on Academic Peer Reviews

Large language models LLMs are increasingly integrated into academic workflows, with many conferences and journals permitting their use for tasks such as language refinement and literature summarization. However, their use in peer review remains prohibited due to concerns around confidentiality...

Phare: a Safety Probe for Large Language Models

Ensuring the safety of large language models LLMs is critical for responsible deployment, yet existing evaluations often prioritize performance over identifying failure modes. We introduce Phare, a multilingual diagnostic framework to probe and evaluate LLM behavior across three critical...

A Survey on the Safety and Security Threats of Computer-Using Agents: JARVIS or Ultron?

Recently, AI-driven interactions with computing devices have advanced from basic prototype tools to sophisticated, LLM-based systems that emulate human-like operations in graphical user interfaces. We are now witnessing the emergence of \emphComputer-Using Agents CUAs, capable of autonomously...

Evaluating Query Efficiency and Accuracy of Transfer Learning-Based Model Extraction Attack in Federated Learning

Federated Learning FL is a collaborative learning framework designed to protect client data, yet it remains highly vulnerable to Intellectual Property IP threats. Model extraction ME attacks pose a significant risk to Machine Learning as a Service MLaaS platforms, enabling attackers to replicate...

CVE-2024-32646

Vyper is a pythonic Smart Contract Language for the Ethereum virtual machine. In versions 0.3.10 and prior, using the slice builtin can result in a double eval vulnerability when the buffer argument is either msg.data, self.code or .code and either the start or length arguments have side-effects...

CVE-2024-33327

A cross-site scripting XSS vulnerability in the component UrlAccessibilityEvaluation.jsp of Lumisxp v15.0.x to v16.1.x allows attackers to execute arbitrary web scripts or HTML via a crafted payload injected into the contentHtml parameter...

CVE-2024-35229

ZKsync Era is a layer 2 rollup that uses zero-knowledge proofs to scale Ethereum. Prior to version 1.3.10, there is a very specific pattern fa,b; checkifaexecutedlast in Yul that exposes a bug in evaluation order of Yul function arguments. This vulnerability has been fixed in version 1.3.10. As a...

CVE-2024-32647

Vyper is a pythonic Smart Contract Language for the Ethereum virtual machine. In versions 0.3.10 and prior, using the createfromblueprint builtin can result in a double eval vulnerability when rawargs=True and the args argument has side-effects. It can be seen that the buildcreateIR function of t...

CVE-2024-32649

Vyper is a pythonic Smart Contract Language for the Ethereum virtual machine. In versions 0.3.10 and prior, using the sqrt builtin can result in double eval vulnerability when the argument has side-effects. It can be seen that the buildIR function of the sqrt builtin doesn't cache the argument to...

CVE-2023-31032

NVIDIA DGX A100 SBIOS contains a vulnerability where a user may cause a dynamic variable evaluation by local access. A successful exploit of this vulnerability may lead to denial of service...

Arbitrary Code Execution

Langroid is vulnerable to Arbitrary Code Execution. The vulnerability is due to unsafe code evaluation due to the use of pandas.eval in the LanceDocChatAgent via the computefromdocs function, allowing attackers to execute malicious code through unsanitized input...

CVE-2023-41052

Vyper is a Pythonic Smart Contract Language. In affected versions the order of evaluation of the arguments of the builtin functions uint256addmod, uint256mulmod, ecadd and ecmul does not follow source order. This behaviour is problematic when the evaluation of one of the arguments produces side...

CVE-2023-36118

Cross Site Scripting vulnerability in Faculty Evaulation System using PHP/MySQLi v.1.0 allows an attacker to execute arbitrary code via a crafted payload to the page parameter...

CVE-2023-33569

Sourcecodester Faculty Evaluation System v1.0 is vulnerable to arbitrary code execution via ip/eval/ajax.php?action=updateuser...

CVE-2023-33439

Sourcecodester Faculty Evaluation System v1.0 is vulnerable to SQL Injection via /eval/admin/managetask.php?id=...

CVE-2023-31844

Sourcecodester Faculty Evaluation System v1.0 is vulnerable to SQL Injection via /eval/admin/managesubject.php?id=...

CVE-2023-2369

A vulnerability was found in SourceCodester Faculty Evaluation System 1.0. It has been rated as critical. This issue affects some unknown processing of the file admin/managerestriction.php. The manipulation of the argument id leads to sql injection. The attack may be initiated remotely. The explo...