Academic papers, new techniques, benchmarks, and theoretical findings in AI/LLM security.
This academic paper proposes a policy-based conjunctive scheme, which is a method for managing how groups of people can collectively decide to delete shared data they all own together. The research addresses the challenge of 'digital forgetting' (the ability to have data permanently removed) when multiple parties have rights to the same information, requiring agreement from all co-owners before deletion occurs.
This academic paper, published in July 2026, presents research on laconic attribute-based PSI (private set intersection, a technique that lets two parties find common items in their datasets without revealing the full datasets to each other) applied to authenticated inputs. The work appears to focus on theoretical cryptographic methods for secure data comparison while maintaining privacy and verifying that the data being compared is legitimate.
This research paper examines how well attack mitigations (security protections built into code) actually work in Rust and Go, two programming languages designed to be memory-safe (meaning they prevent common memory-related bugs that attackers often exploit). The study analyzes real compiled programs to see whether these language protections hold up against real-world attacks.
This research paper examines how people with and without prior victimization differ in their ability to detect scams. The study, published in Computers & Security in May 2026, explores whether past experience being scammed makes individuals better at identifying fraudulent attempts.
This research paper presents MPV, a method for restricting access to master keys in multi-user Paillier systems (a cryptographic system that allows certain calculations on encrypted data without decrypting it first) by using mixed ciphertexts (encrypted data created with different encryption methods combined). The approach aims to improve security by preventing unauthorized parties from decrypting sensitive information even if they gain access to the master key.
AI agents (software systems that can plan and take actions over time) that retain memory between sessions create a security risk called Memory & Context Poisoning, where attackers can inject malicious instructions into persistent storage that the agent continues to trust and follow in future interactions. Researchers found a vulnerability called MemoryTrap in Claude Code where a developer could unknowingly approve a malicious dependency that would persist in the agent's memory and configuration files, poisoning the agent's behavior across multiple projects and sessions. The core problem is that agents treat stored memory, configuration files, and hooks as trustworthy guidance without validating whether they contain attacker-controlled content.
Fix: Anthropic released Claude Code v2.1.50, which removed user memories from the system prompt (the core instructions that guide the AI's behavior) to reduce the specific attack path that MemoryTrap exploited.
OWASP GenAI SecurityThis academic publication discusses PUF (physically unclonable functions, unique fingerprints built into hardware chips that are nearly impossible to copy) optimization methods for authenticating IoT devices (internet-connected devices like smart home sensors). The research focuses on improving how these hardware-based security features can be used to verify that IoT devices are genuine and trustworthy.
This academic paper presents a method using AI to extract entities (named items like organizations or IP addresses) and relationships between them from threat intelligence data about APT (advanced persistent threat, a type of sophisticated cyberattack) attacks. The researchers developed a system to help security analysts automatically identify and organize complex attack patterns from unstructured text documents.
This research paper presents a method for optimally placing honeypots (decoy systems designed to attract and monitor attackers) in networks where multiple attackers operate simultaneously, using Bayesian Stackelberg Games (a mathematical framework for strategic decision-making under incomplete information). The approach aims to help defenders allocate honeypots more effectively by predicting attacker behavior and making strategic placement decisions.
Many students prefer free videos and AI tools over reading security books, even though expert-written books often provide clear and deep knowledge about security. The source encourages students to recognize that security books remain valuable learning resources despite newer alternatives.
Researchers created a hybrid system that combines SAST (static application security testing, which automatically scans code for vulnerabilities) with LLMs (large language models) to better filter and prioritize security alerts. The system reduced false positives (incorrect security warnings) by 91% in real deployments by using AI to intelligently triage findings and generate automated exploit examples.
The General Data Protection Regulation (GDPR, a European law that controls how organizations collect and use personal data) was created to control large tech companies but also applies to smaller organizations like schools. A research study in Italian schools found tension between following strict top-down rules and making practical decisions based on actual risks to protect data.
Evasion attacks (methods where attackers trick AI systems into ignoring safety rules by manipulating input data) have been researched for more than ten years, but most real-world examples remain theoretical and academic. Because these demonstrations seem more like intellectual exercises than practical threats, people have largely dismissed evasion attacks as unimportant in actual security situations.
This research develops a privacy-preserving method for face recognition systems using the Privacy Funnel model, which balances the usefulness of facial data against protecting sensitive information like identity or demographic attributes. The authors introduce new versions of this model, including the Generative Privacy Funnel (GenPF) and deep variational Privacy Funnel (DVPF), and demonstrate that their approach works with modern face recognition systems while reducing information leakage about sensitive attributes.
This paper proposes a new image sharing method that uses compressive sensing (a technique that compresses and encrypts data simultaneously) with multiple privacy levels, so different users can access only the information they need without seeing sensitive details. The method uses an algorithm called T-â„“1-B2DLDA to compress images in a way that allows some users to classify or analyze images without reconstructing the original, while others with higher access levels can fully reconstruct them.
Researchers have developed a dual-branch image tampering detection model that uses two parallel processing paths to identify when images have been altered or forged. The model analyzes both noise patterns (statistical irregularities in pixel data) and anomalous features (unexpected or out-of-place patterns) to detect tampering, offering a more comprehensive approach than methods that examine only one type of indicator.
This research paper examines how machine learning-based network intrusion detection systems (NIDS, software that identifies unauthorized access attempts) can use adaptive active-defense hardening to protect themselves against reinforcement learning (RL, a type of AI that learns by trial-and-error) driven attacks. The study compares this dynamic defense approach with traditional static defenses (fixed security measures that don't change).
This academic paper discusses extending SBOMs (Software Bill of Materials, a detailed list of all software components in a program) to create AIBOMs for agentic AI systems (AI systems that can take independent actions). The research focuses on adding new schema extensions (structured data formats) to track AI agent components, improving how these systems can be orchestrated (coordinated and controlled), and developing methods to evaluate whether AI systems produce consistent, reproducible results.
This research paper proposes a security method for IoV (Internet of Vehicles, where cars connect to networks) that combines blockchain (a distributed ledger technology that creates permanent, tamper-resistant records) with PUF (physical unclonable functions, unique fingerprints derived from hardware that are hard to fake) to create a two-factor authentication and key agreement scheme. The approach aims to improve security for vehicle communication and data exchange in connected car systems.
Researchers proposed DV2PDA, a new method for collecting data from Industrial Internet of Things devices (IIoT, networks of sensors and machines in factories) while protecting privacy and ensuring the data hasn't been tampered with. The scheme uses decentralization (spreading control across multiple computers instead of one central server) and verification (checking that data is authentic) to let organizations aggregate information from many devices without exposing individual sensitive details.