AX·NEWS
Sign in

Cybersecurity

Cybersecurity news — Friday 09 October 2026

124 stories from 9 sources, every one in English. Each links to its publisher.

Open in the live feedRSS feedOn a TV

  1. OpenAI Fires 3 Safety Researchers in Dispute Over AI Risks
    USSecurityWeek

    The ChatGPT maker said the researchers "violated clear policies on handling sensitive information.” The post OpenAI Fires 3 Safety Researchers in Dispute Over AI Risks appeared first on SecurityWeek .

  2. Hackers abuse Google Ads, Bing redirects to push Claude ClickFix attacks
    USBleepingComputer

    Hackers are abusing legitimate Bing search-result redirects as click URLs in Google search ads to direct users to fake Claude installers that deliver ClickFix attacks. [...]

  3. Friday Squid Blogging: I Caught a Squid
    USSchneier on Security

    On Wednesday I spent a day fishing, on a small boat out of Gloucester, MA. We caught many cod (none of which we could keep), and a bunch of hake and mackerel (all of which we could keep). And…I caught a squid! Near as I can tell, it’s a…

  4. Credential-Stealing GitHub Actions Workflows Planted in Tens of Thousands of Repositories
    INThe Hacker News

    Cybersecurity researchers have disclosed details of an ongoing credential-theft campaign that has compromised two high-profile open-source maintainer accounts to push a malicious workflow into over 340 repositories. "Using the account of…

  5. 4 ways Amazon’s new Alexa tablets eclipse the Fire line – starting with Google Play
    USZDNet Security

    Amazon just upped its game with new Android tablets for entertainment and work.

  6. ‘I use AI to do the things that I’m bad at’: Linus Torvalds on why it works for him
    USZDNet Security

    The Linux creator calls vibe coding ‘a wonderful way to find joy in programming,’ but not for running Linux development. (Finding bugs is OK.)

  7. FBI Arrests Another ShinyHunters Suspect Reportedly Involved in Its Jobs Portal Hack
    INThe Hacker News

    The FBI has arrested another suspected co-conspirator of ShinyHunters, FBI Director Kash Patel said on October 9 in a post on X. ShinyHunters is the extortion group that said in September it had breached the FBI's jobs portal and stolen…

  8. Unpatched AhsayCBS flaws exploited to deploy webshells, mine crypto
    USBleepingComputer

    Threat actors are exploiting one critical and one medium-severity vulnerability still unpatched in the AhsayCBS backup management platform to deploy webshells and cryptocurrency miners. [...]

  9. FBI arrests another suspected ShinyHunters hacker after agency breach
    USBleepingComputer

    The FBI has arrested another suspected member of the ShinyHunters extortion group believed to be involved in the recent breach of FBI systems, Director Kash Patel announced Friday. [...]

  10. P7 DarkSword iOS Exploit Kit Adds Crypto Wallet Data Theft and Remote Commands
    INThe Hacker News

    Cybersecurity researchers have disclosed details of a previously unseen variant of the DarkSword iOS exploit kit called P7 DarkSword. "Compared with the variants we usually observe, P7 reduces its on-device footprint, adds on-device…

  11. Germany arrests alleged core Qilin ransomware member after extradition
    USBleepingComputer

    Germany has arrested a Russian national suspected of being a leading member of the Qilin ransomware group following extradition from Japan earlier this month. [...]

  12. How to keep AI agents within their permissions
    USBleepingComputer

    AI agents can use valid credentials to perform actions beyond their assigned permissions, creating risks that traditional access controls may not prevent. Token Security explains how organizations can enforce agent-specific policies…

  13. TP-Link Sued by Four More U.S. States Over Router Security and China Ties
    INThe Hacker News

    Four more U.S. states sued router maker TP-Link Systems on October 6, bringing the total to five, with Texas filing a suit in February. Florida, Iowa, Montana and Nebraska allege the California company misled buyers about how secure its…

  14. Researchers Publish Working Exploit for Pre-Auth AnyDesk Linux Flaw That Gives Root Access
    INThe Hacker News

    Security researchers have published a full working exploit for a pre-authentication remote code execution flaw in AnyDesk Linux that gives attackers root access before anyone approves the connection. AnyDesk patched the flaw in version…

  15. Anthropic Launches Free AI Vulnerability Scanner for Open-Source Projects
    INThe Hacker News

    Anthropic on Thursday unveiled OSS Scanner as an opt-in vulnerability scanner to help secure the open-source ecosystem using artificial intelligence (AI). "It's an opt-in service informed by our experience using Claude to find…

  16. Attackers Exploit AhsayCBS Flaws to Deploy XMRig Miners Disguised as Microsoft Edge
    INThe Hacker News

    Threat actors have been observed exploiting two recently disclosed flaws in the AhsayCBS backup utility to seize control of affected devices and deploy web shells and XMRig cryptocurrency miners. Details of the flaws are below …

  17. Max severity SonicWall SMA1000 flaw now exploited in attacks
    USBleepingComputer

    Attackers are exploiting a maximum-severity vulnerability in SonicWall SMA1000 appliances (CVE-2026-102255) that was patched on Tuesday, three days ago. [...]

  18. Flax Typhoon Exploits Five Flaws as CISA Sets October 11 Deadline for Federal Agencies
    INThe Hacker News

    The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Thursday added five security flaws to its Known Exploited Vulnerabilities (KEV) catalog, following their abuse by a China-linked threat actor known as Flax Typhoon. The…

  19. In Other News: AI Used in Korean Bank Breaches, Poem-Guided Botnet, Empire Admin Gets 40 Years
    USSecurityWeek

    Noteworthy stories that might have slipped under the radar: Tensorlake npm SDK compromised, Empire Market co-founder gets 40 years, exposed NVIDIA GPU monitors leak telemetry. The post In Other News: AI Used in Korean Bank Breaches…

  20. Google Domains Impacted by Recent ccTLD Hijacks
    USSecurityWeek

    Hackers hijacked the .gh, .sl, and .as ccTLDs and obtained HTTPS certificates for several Google domains. The post Google Domains Impacted by Recent ccTLD Hijacks appeared first on SecurityWeek .

  21. The AI Velocity Paradox: Why Security Is Decades Behind AI Ambition
    INThe Hacker News

    As enterprises race to deploy autonomous AI agents to accelerate business, a new report reveals they are tethered to security architectures built for a different era. The "Horizons of Identity Security" report from SailPoint highlights a…

  22. Man admits to running network of 15,000 money mules for cybercriminals
    USBleepingComputer

    ​A Ukrainian-Russian dual citizen has pleaded guilty to running a massive money laundering operation that laundered millions for cybercriminals worldwide. [...]

  23. ASOS breach update: Hackers stole customer details and shopping searches
    USMalwarebytes Labs

    The customer data stolen from global fashion retailer ASOS includes more than just names and contact details, raising questions about its early reassurances. As we reported earlier this week, ASOS customers received a push notification…

  24. Unpatched AhsayCBS Vulnerabilities Exploited in the Wild
    USSecurityWeek

    The flaws, CVE-2026-105133 and CVE-2026-105134, allow attackers to bypass authentication and inject OS commands. The post Unpatched AhsayCBS Vulnerabilities Exploited in the Wild appeared first on SecurityWeek .

  25. Microsoft: Outdated Windows devices will stop receiving security updates
    USBleepingComputer

    Microsoft says devices running unsupported versions of Windows will stop receiving security updates after next year's Windows Update certificate rotation. [...]

  26. Pre-Baked Firmware Malware Hits Budget Android Devices in 150+ Countries
    USSecurityWeek

    Midnight Mimosa is the name given to a malware campaign primarily running preinstalled on low-cost Android devices. The post Pre-Baked Firmware Malware Hits Budget Android Devices in 150+ Countries appeared first on SecurityWeek .

  27. GoBalance Flaw Lets Attackers Hijack .onion Addresses by Recovering Tor-Format Keys
    INThe Hacker News

    A bug in GoBalance, a tool many dark-web sites use to stay reachable during attacks, lets anyone work out the secret key that controls a site's .onion address using only public information, and then take that address over. Searchlight…

  28. US Disrupts Chinese State-Sponsored Hacking Tools
    USSecurityWeek

    Flax Typhoon and other APTs used MicroScan and FishHub to scan and hack US and foreign critical infrastructure. The post US Disrupts Chinese State-Sponsored Hacking Tools appeared first on SecurityWeek .

  29. Citrix warns admins to patch new NetScaler RCE flaw immediately
    USBleepingComputer

    Citrix has warned IT administrators to patch systems immediately against a new critical vulnerability affecting NetScaler ADC networking appliances and NetScaler Gateway secure remote access solutions. [...]

  30. Three Teams Demonstrate Remote Hacks of Fully Patched Google Pixel 10 at Pwn2Own
    INThe Hacker News

    Three research teams broke into Google's Pixel 10 on October 8 at Pwn2Own Ireland, a hacking contest in Cork whose rules require every target to be fully patched. The contest pays researchers to show working exploits and passes the flaws…

  31. Anthropic Fast-Tracks AI Bug Reports to OSS Maintainers, Taps 11 Firms for OT Security
    USSecurityWeek

    OSS Scanner sends unreviewed, model-generated vulnerability reports to open source maintainers that opt in. The post Anthropic Fast-Tracks AI Bug Reports to OSS Maintainers, Taps 11 Firms for OT Security appeared first on SecurityWeek .

  32. Citrix Patches Critical NetScaler Flaw That Could Enable RCE in SAML Deployments
    INThe Hacker News

    Citrix has released patches for yet another critical security flaw impacting NetScaler ADC and NetScaler Gateway that could result in remote code execution or denial-of-service (DoS) under certain conditions. "CVE-2026-107406 is a memory…

  33. ISC Stormcast For Friday, October 9th, 2026 https://isc.sans.edu/podcastdetail/10130, (Fri, Oct 9th)
    USSANS ISC
  34. Citrix Urges Immediate Patching of Critical NetScaler Vulnerability
    USSecurityWeek

    The security defect, tracked as CVE-2026-107406, could lead to remote code execution or denial-of-service. The post Citrix Urges Immediate Patching of Critical NetScaler Vulnerability appeared first on SecurityWeek .

  35. Google Pixel 10 Exploits Earned Hackers $560,000 at Pwn2Own
    USSecurityWeek

    $1.2 million was paid out at Pwn2Own Ireland 2026 for exploits targeting phones, printers, smart speakers, smart home hubs, and AI infrastructure and coding tools. The post Google Pixel 10 Exploits Earned Hackers $560,000 at Pwn2Own…

  36. FBI Seizes 7 Domains, Disrupts Flax Typhoon Tools Used in Critical Infrastructure Intrusions
    INThe Hacker News

    The U.S. Federal Bureau of Investigation (FBI) and Department of Justice (DoJ) have announced the disruption of malicious tools used by a China-linked advanced persistent threat group known as Flax Typhoon. To that end, the agencies seized…

  37. Hackers get $1,262,000 for 98 zero-days at Pwn2Own Ireland
    USBleepingComputer

    The Pwn2Own Ireland 2026 hacking contest has concluded, with hackers collecting $1,262,000 in rewards after exploiting 98 zero-day flaws. [...]

  38. Formula Predicts When AI Chatbots Are at Risk of Turning Bad
    USSecurityWeek

    Researchers from George Washington University have published a paper examining whether the time and cause of AI going rogue can be predicted. The post Formula Predicts When AI Chatbots Are at Risk of Turning Bad appeared first on…

  39. OTRO: Oblivious Tokenization Path with Square-Root ORAM
    USarXiv cs.CR (Security)

    arXiv:2606.17358v3 Announce Type: replace Abstract: The CPU-side large language model (LLM) tokenizer is a critical security gap in LLM serving through a confidential computing stack with CPU and GPU trusted execution environments (TEEs)…

  40. An Autonomous AI Security Agent for Banking: Multi-Vector Fraud and AML Detection Across Retail and Corporate Accounts
    USarXiv cs.CR (Security)

    arXiv:2606.17555v4 Announce Type: replace Abstract: Banks face two threat families with fundamentally different detection requirements: signature-based fraud (card-not-present attacks, account takeover, ATM cloning) and behavioural…

  41. ShannonProver: Towards Automating Formal Cryptographic Proofs
    USarXiv cs.CR (Security)

    arXiv:2607.02847v4 Announce Type: replace Abstract: Cryptographic proofs are produced at a scale that increasingly exceeds the community's ability to verify them manually. Machine-checked proofs offer a path toward scalable proof…

  42. Balancing Privacy and Compliance in DeFi: A Zero-Knowledge-Based Auditable Cross-Chain Framework
    USarXiv cs.CR (Security)

    arXiv:2608.15276v2 Announce Type: replace Abstract: The rapid emergence of decentralized finance (DeFi) has introduced complex challenges for cross-chain transactions, which involve transferring assets across disparate blockchain networks…

  43. Evaluating and Improving the Robustness of Large Language Models to Input Sequence Variations
    USarXiv cs.CR (Security)

    arXiv:2610.02432v2 Announce Type: replace Abstract: Large language models (LLMs) in production systems face prompt injections, trojans (backdoors), and manipulation of automatic quality metrics. This thesis develops models, methods, and…

  44. IPEK: Event-Criticality-Aware Evidential Trust Management against Strategic Insider Attacks in Vehicular Ad Hoc Networks
    USarXiv cs.CR (Security)

    arXiv:2604.07532v3 Announce Type: replace-cross Abstract: Event-based trust management in vehicular ad hoc networks is vulnerable to strategic insider attackers who report trivial events correctly to build reputation, falsify reports on…

  45. From Reconnaissance to Response: Quantitative Risk Parameterization and Game Theoretic Containment in Modern Enterprise Attack
    USarXiv cs.CR (Security)

    arXiv:2609.36485v2 Announce Type: replace-cross Abstract: Modern Security Operations Centers struggle with delayed manual incident response, enabling adversaries to advance through the Cyber Kill Chain during early stage reconnaissance…

  46. One-Shot Private Confidence Regions via Resampling
    USarXiv cs.CR (Security)

    arXiv:2610.08460v2 Announce Type: replace-cross Abstract: We propose a simple framework for constructing differentially private confidence regions \textit{in one shot}, i.e., by adding noise only to the final resampling quantile instead of…

  47. SoK: Failure Modes in Common Criteria Product Evaluation - A Taxonomy and Design-for-Evaluability Guidance
    USarXiv cs.CR (Security)

    arXiv:2610.10644v1 Announce Type: new Abstract: The Common Criteria (CC; ISO/IEC 15408) is the principal international framework for evaluating the security of IT products, and its certificates gate procurement across government, defense…

  48. BRACE: Differential Privacy for Dense Associative Memory with LSR Energy
    USarXiv cs.CR (Security)

    arXiv:2610.11218v1 Announce Type: new Abstract: Dense associative memory (DAM) provides an energy-based framework for memory retrieval with close connections to attention mechanisms in modern artificial intelligence. Despite growing…

  49. MORDOR:Mitigating Overheads of Read Disturbance Preventive Operations via Elastic Refresh Scheduling
    USarXiv cs.CR (Security)

    arXiv:2610.11398v1 Announce Type: new Abstract: Modern DRAM chips are susceptible to read disturbance phenomena such as RowHammer, where repeatedly accessing (hammering) a row of DRAM cells (i.e., a DRAM row) induces bitflips in other…

  50. Accelerating HQC for Post-Quantum TLS 1.3 on x86 IoT Gateways
    USarXiv cs.CR (Security)

    arXiv:2610.11107v1 Announce Type: new Abstract: Post-quantum TLS at an IoT gateway must protect many device connections without large increases in handshake delay, CPU cost, or network traffic. HQC provides code-based diversity beyond…

  51. ReSI: Recursive Safety Improvement toward Resistant and Resilient AI
    USarXiv cs.CR (Security)

    arXiv:2610.12233v1 Announce Type: new Abstract: Recursive self-improvement, the participation of AI systems in improving their own capabilities, is beginning to move from theoretical prospect to practice, posing both challenges and…

  52. Safe at One Loop, Risky at Another: Aligning Safety Across Recurrent Depths in Looped Language Models
    USarXiv cs.CR (Security)

    arXiv:2610.10625v1 Announce Type: new Abstract: Looped Language Models (LoopLMs) provide a parameter efficient approach to scaling model capabilities through repeated use of shared parameters across recurrent steps. Since each recurrent…

  53. The Sample Complexity of Membership Inference and Privacy Auditing
    USarXiv cs.CR (Security)

    arXiv:2508.19458v2 Announce Type: replace-cross Abstract: A membership-inference attack gets the output of a learning algorithm, and a target individual, and tries to determine whether this individual is a member of the training data or an…

  54. Beyond Direct Access: Resource Hijacking in LLM Agents
    USarXiv cs.CR (Security)

    arXiv:2608.15108v2 Announce Type: replace Abstract: Large language model agents are increasingly connected to high-value resources, including external APIs, GPUs and servers, and workflows such as deployment and approval. Existing agent…

  55. Detection-Guided Adaptive Purification with Diffusion Models for Robust Audio Deepfake Detection
    USarXiv cs.CR (Security)

    arXiv:2610.10752v1 Announce Type: new Abstract: Audio deepfake detectors remain vulnerable to adversarial perturbations that suppress the acoustic cues used for detection, allowing manipulated utterances to evade the detector. Although…

  56. Black-Box Forensics for Conversational LLM Agents
    USarXiv cs.CR (Security)

    arXiv:2606.22698v2 Announce Type: replace Abstract: As LLM-powered scams proliferate, black-box forensics for conversational LLM agents offers a path to accountability for systems hidden behind anonymous endpoints. Identifying the base…

  57. Arbiter: Detecting Interference in LLM Agent System Prompts
    USarXiv cs.CR (Security)

    arXiv:2603.08993v2 Announce Type: replace-cross Abstract: System prompts for LLM-based coding agents are software artifacts that govern agent behavior, yet lack the testing infrastructure applied to conventional software. We present…

  58. Power Side-Channel Membership Inference Attack on Embedded Machine Learning
    USarXiv cs.CR (Security)

    arXiv:2610.10909v1 Announce Type: new Abstract: Membership inference attacks (MIAs) threaten the privacy of machine learning (ML) training data by determining whether a sample was used to train a target model. Existing MIAs rely on model…

  59. When AI Finds Hidden Messages, Does It Report?
    USarXiv cs.CR (Security)

    arXiv:2610.10620v1 Announce Type: new Abstract: When an assistant encounters a message for another AI, does it tell its user? Four fixed model-provider deployments perform simulated source tasks in 1,280 ordinary-note and 128 enhanced-note…

  60. ORCAGen: Orchestrating Context-Aware Malware Deception with RAG-Guided Generative AI
    USarXiv cs.CR (Security)

    arXiv:2610.12415v1 Announce Type: new Abstract: Malware defenses often remove or isolate suspicious programs as quickly as possible. While effective for containment, this approach can also waste an opportunity to observe attacker behavior…

  61. The Matrix Subcode Equivalence problem and its application to signature with MPC-in-the-Head
    USarXiv cs.CR (Security)

    arXiv:2507.15377v2 Announce Type: replace Abstract: Nowadays, equivalence problems are widely used in cryptography, most notably to establish cryptosystems such as digital signatures, with MEDS, LESS, PERK as the most recent ones. However…

  62. PackMonitor: Enabling Zero Package Hallucinations Through Decoding-Time Monitoring
    USarXiv cs.CR (Security)

    arXiv:2602.20717v2 Announce Type: replace-cross Abstract: As Large Language Models (LLMs) are increasingly integrated into software development workflows, their trustworthiness has become a critical concern. However, in dependency…

  63. Language Modeling is Monotone Compression
    USarXiv cs.CR (Security)

    arXiv:2610.11031v1 Announce Type: cross Abstract: A long-standing hypothesis in artificial intelligence and neuroscience posits that intelligence is closely related to compression: the ability to compress information efficiently…

  64. When Routing Reveals Membership: Privacy Leakage from MoE Router Telemetry
    USarXiv cs.CR (Security)

    arXiv:2610.10616v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) language models produce routing information during inference that may be logged or exposed for monitoring, debugging, load analysis, and safety auditing. Unlike…

  65. Applying Security by Design at the Point of Execution: How Governed Security Requirements Affect the Security of AI-Generated Code
    USarXiv cs.CR (Security)

    arXiv:2610.10659v1 Announce Type: cross Abstract: Security by design asks that security requirements are defined before code is written. Secure-code benchmarks typically measure the opposite situation: the agent receives the task without…

  66. Anytime-valid detection of LLM weight exfiltration
    USarXiv cs.CR (Security)

    arXiv:2610.11843v1 Announce Type: new Abstract: A compromised LLM inference server can leak model weights by encoding payload bits in otherwise plausible token choices. A replay of the same prompt in a trusted server can expose such…

  67. From Public Posts to AI-Search Citations: Measuring the Fragility of AI Search
    USarXiv cs.CR (Security)

    arXiv:2610.11932v1 Announce Type: new Abstract: As more users ask AI systems for information, AI-search platforms are becoming a common gateway to web information. Unlike traditional search, which maps keywords to ranked pages, AI search…

  68. PyCache Trap: The Inspection-Execution Gap in Agent Skill Scanners
    USarXiv cs.CR (Security)

    arXiv:2610.10612v1 Announce Type: new Abstract: Agent skills combine instructions with executable resources, giving third-party packages access to an agent's runtime. Existing skill scanners inspect documentation and visible source, but…

  69. Speedbumps: Rejection Attacks on Speculative Decoding
    USarXiv cs.CR (Security)

    arXiv:2610.10929v1 Announce Type: new Abstract: Speculative decoding is a popular technique for increasing the speed and reducing the costs of large language model (LLM) inference by verifying multiple draft tokens in a single target-model…

  70. MARC: Multi-Bit Watermarking for Autoregressive Audio Generation against Codec Attacks
    USarXiv cs.CR (Security)

    arXiv:2610.11488v1 Announce Type: new Abstract: Generated audio is now used in a range of applications, creating a need to verify its origin after distribution and signal processing. This task is particularly challenging for autoregressive…

  71. A Security Meta-Model for Retrieval-Augmented Generation Systems
    USarXiv cs.CR (Security)

    arXiv:2610.11893v1 Announce Type: new Abstract: Retrieval-Augmented Generation (RAG) systems extend large language models (LLMs) with external knowledge through a multi-stage pipeline. While this architecture can improve the factual…

  72. LTBD: Learnable Trust-Boundary Delimiters for Prompt Injection Defense
    USarXiv cs.CR (Security)

    arXiv:2610.11634v1 Announce Type: new Abstract: Large language models (LLMs) perform remarkably well on complex tasks, yet remain highly vulnerable to prompt injection attacks, where malicious instructions embedded in external data can…

  73. CPU-Auth: Device Fingerprinting for Authentication via DVFS Side-Channel
    USarXiv cs.CR (Security)

    arXiv:2610.10766v1 Announce Type: new Abstract: Lack of effective authentication has resulted in numerous security and privacy breaches, including unauthorized access to protected information, identity theft, and fraud. One approach to…

  74. ProxyEraseAgent: Blind Watermark Removal in the Wild
    USarXiv cs.CR (Security)

    arXiv:2610.11290v1 Announce Type: new Abstract: Invisible image watermark removal has received growing attention. Despite substantial progress, existing attacks face a tension between practicality and specificity. Attacks exploiting…

  75. Characterizing Statistical Separability in TP-CRIV for Probabilistic AI Models
    USarXiv cs.CR (Security)

    arXiv:2610.11163v1 Announce Type: new Abstract: Third-party challenge-response identity verification (TP-CRIV) enables an independent verifier to assess whether a claimant possesses a model identical to a remotely deployed model without…

  76. SoK: Are LLMs Reliable at Source Code Recovery? A Taxonomy and Empirical Evaluation
    USarXiv cs.CR (Security)

    arXiv:2610.11556v1 Announce Type: new Abstract: Effective source recovery is critical to security applications such as malware analysis, vulnerability assessment, and legacy maintenance. Large Language Models (LLMs) are reshaping this…

  77. SemField: A Simple, Linear, Continuous, yet Robust Semantic Watermark
    USarXiv cs.CR (Security)

    arXiv:2610.11848v1 Announce Type: new Abstract: The rapid proliferation of Large Language Models (LLMs) necessitates reliable watermarking techniques to identify AI-generated text and ensure appropriate attribution. While token-based…

  78. On-Chain Archaeology of Bitcoin Oracles: Evidence of Use under Limited Observability
    USarXiv cs.CR (Security)

    arXiv:2610.11439v1 Announce Type: new Abstract: Before Ethereum made the "oracle problem" a household term, Bitcoin already had oracles serving as feeds, key-release services, federated signers, and arbiters that carried real value on the…

  79. BRANCH: Bypassing Multi-Scanner AI Guardrails
    USarXiv cs.CR (Security)

    arXiv:2610.10742v1 Announce Type: new Abstract: AI systems increasingly rely on Large Language Models (LLMs) as core reasoning engines, making them targets for prompt injection and jailbreaks. Guardrails monitor and validate model inputs…

  80. Certified Corruption Budgets: Anytime-Valid Leaderboard Claims under Adaptive Rigging
    USarXiv cs.CR (Security)

    arXiv:2610.10597v1 Announce Type: new Abstract: Public leaderboards for AI models are read continuously, and attackers can see every published standing. Vote rigging, selective disclosure of private variants, and benchmark contamination…

  81. Improved Local Leakage Resilience of Shamir Secret Sharing and Worst-Case Optimal Polynomial Intersection
    USarXiv cs.CR (Security)

    arXiv:2610.12357v1 Announce Type: cross Abstract: We study two problems: Local Leakage Resilience (LLR) for Shamir secret sharing, and worst-case Optimal Polynomial Intersection (OPI). Both problems concern polynomials $Q(X)$ of degree…

  82. Adversarial Cues in Decision Models Used as Judges: The Role of Request Presentation
    USarXiv cs.CR (Security)

    arXiv:2610.11436v1 Announce Type: cross Abstract: An answer judge instructed to grade the final commitment should reject an explicitly wrong final value even when an earlier value matches the reference. We show that adding one colon to a…

  83. Verification with Transfer: Exact Information Frontiers and Their Price in Calls
    USarXiv cs.CR (Security)

    arXiv:2610.12211v1 Announce Type: cross Abstract: A verifier that accepts or rejects whole answers reveals little: under a flat prior over $k$-bit answers, zero error needs $2^k-1$ verifications. The usual remedy is to solve related source…

  84. Proof-of-Continuity: A Temporal Model for Authority Propagation in Distributed Systems and AI Agents
    USarXiv cs.CR (Security)

    arXiv:2607.08906v2 Announce Type: replace Abstract: Proof-of-Possession authorization models derive authority from the possession of artifacts such as tokens, credentials, or capabilities. This paper argues that possession is insufficient…

  85. Certifying Hidden Paths: Scalable Topology Assurance for QKD Networks
    USarXiv cs.CR (Security)

    arXiv:2610.11513v1 Announce Type: new Abstract: Large-scale Quantum Key Distribution (QKD) networks rely on trusted repeaters, making the security properties of the selected communication path an essential part of end-to-end assurance. At…

  86. DCVD: Dual-Channel Cross-Modal Fusion for Joint Vulnerability Detection and Localization
    USarXiv cs.CR (Security)

    arXiv:2605.11015v2 Announce Type: replace Abstract: Software vulnerability detection plays a critical role in ensuring system security, where real-world auditing requires not only determining whether a function is vulnerable but also…

  87. Pretext: Defeating Malicious Skill Detection Frameworks for AI Agents
    USarXiv cs.CR (Security)

    arXiv:2609.39607v2 Announce Type: replace Abstract: Skills extend an agent's capabilities by injecting instructions and information into the context, and are widely used by agents such as OpenClaw and Claude Code. Prior work shows…

  88. Soft Voting for Policy-Aware Private Data Synthesis
    USarXiv cs.CR (Security)

    arXiv:2610.11285v1 Announce Type: new Abstract: Blowfish privacy relaxes differential privacy (DP) by protecting only the attribute-value substitutions a data owner specifies as edges of a policy graph. A sparser policy can reduce the…

  89. Defensive Sufficiency in a Stackelberg Model of AI Security
    USarXiv cs.CR (Security)

    arXiv:2610.09892v2 Announce Type: replace Abstract: Feedback from automated testing, human red teaming, and incident response can strengthen an AI system's defenses when discovered failures lead to effective repairs. We study when this…

  90. Nullify: Null-Space Activation Steering for Training-Free LLM Unlearning
    USarXiv cs.CR (Security)

    arXiv:2610.10655v1 Announce Type: cross Abstract: Large Language Models (LLMs) inevitably internalize substantial amounts of sensitive or private information during pre-training, while LLM unlearning aims to selectively erase specific…

  91. Provable Subexponential Algorithms for NIST Third-Round Lattice Families
    USarXiv cs.CR (Security)

    arXiv:2610.11254v1 Announce Type: new Abstract: We give provable classical subexponential algorithms for secret recovery in growing parameter families associated with NIST third-round lattice candidates. For the Kyber/ML-KEM, FrodoKEM…

  92. EIFL: Efficiently Protecting Global Model Privacy and Integrity Against an Untrusted Server in Federated Learning
    USarXiv cs.CR (Security)

    arXiv:2610.11511v1 Announce Type: new Abstract: Federated learning (FL) typically adopts a server-client architecture, where the server aggregates clients' local models (i.e., the input) and returns the aggregated global model (i.e., the…

  93. DITTO: A Context-aware Pickle-based Pre-Trained Model Scanner for Effective Security Audits
    USarXiv cs.CR (Security)

    arXiv:2610.10735v1 Announce Type: new Abstract: Pre-trained models (PTMs) are widely distributed as serialized binaries, but their reuse often exposes software supply chains to deserialization attacks. Despite the emergence of safer…

  94. A Zero-Knowledge Signature Framework for Efficient Post-Quantum Message Authentication in Cooperative Automated Driving
    USarXiv cs.CR (Security)

    arXiv:2610.11490v1 Announce Type: new Abstract: Connected and Automated Vehicles (CAV) rely on authenticated Vehicle-to-Everything (V2X) communications to exchange safety-critical information among vehicles and roadside infrastructure. As…

  95. Dynamic Free-Rider Detection in Cross-Silo Federated Learning via Simulated Attack Patterns
    USarXiv cs.CR (Security)

    arXiv:2604.04611v3 Announce Type: replace-cross Abstract: Federated learning (FL) enables multiple clients to collaboratively train a global model by aggregating local updates without sharing private data. In this work, we focus on…

  96. MRCert: Towards Post-deployment Patch Robustness Certification for Adversarially Patched Samples via Type-specific Masking
    USarXiv cs.CR (Security)

    arXiv:2610.10617v1 Announce Type: new Abstract: In post-deployment time, inputs to deep learning models may or may not be adversarially patched. Patch robustness certification on such inputs within a patch bound can verify their label…

  97. Temporal transformer CAN encoder with federated lightweight heads for anomaly detection
    USarXiv cs.CR (Security)

    arXiv:2610.10613v1 Announce Type: cross Abstract: Modern vehicles rely on large numbers of Electronic Control Units (ECUs) that constantly exchange information over the Controller Area Network (CAN) bus. Due to the rapidity, structure, and…

  98. mAVE: A Watermark for Joint Audio-Visual Generation Models
    USarXiv cs.CR (Security)

    arXiv:2603.07090v2 Announce Type: replace Abstract: Watermarking joint audio-visual generation supports vendor copyright protection and content provenance. However, independently valid audio and video watermarks do not establish a shared…

  99. When Flaws Cascade: Understanding Vulnerabilities and Exploitation Chains in JavaScript Engines
    USarXiv cs.CR (Security)

    arXiv:2610.10844v1 Announce Type: new Abstract: JavaScript engines are pivotal to modern web browsers, enabling the execution of dynamic and interactive web applications. However, their complexity and widespread adoption make them prime…

  100. Understanding and Enhancing Backdoor Persistency in LLM Agent Post-Training
    USarXiv cs.CR (Security)

    arXiv:2610.07510v2 Announce Type: replace Abstract: Developers can build LLM agents by adapting third-party models through benign post-training. We study a supply-chain threat in which an attacker supplies a model with a backdoor: hidden…

  101. From Investigation Failures to Reliable SOC Agents: Understanding and Improving LLM-Based Alert Triage
    USarXiv cs.CR (Security)

    arXiv:2610.10608v1 Announce Type: new Abstract: Security operations centers (SOCs) must triage large volumes of alerts, most of which are benign, while missed attacks can remain uninvestigated. Tool-using large language model (LLM) agents…

  102. Could LLM Watermark Detection be Public?
    USarXiv cs.CR (Security)

    arXiv:2610.12106v1 Announce Type: new Abstract: Watermarking large language models is popular for tracing chatbot and agentic outputs, yet detectors remain unreleased since exposing them could let attackers do targeted edits with the…

  103. GROB: A Multi-Agent Architecture for Public-Trace Investigation of Candidate Agentic Activity
    USarXiv cs.CR (Security)

    arXiv:2610.11467v1 Announce Type: new Abstract: We present GROB, a multi-agent architecture for investigating candidate autonomous-agent activity through public Internet traces when privileged telemetry is unavailable. The system performs…

  104. Token by Token, Compromised: Backdoor Vulnerabilities in Unified Autoregressive Models
    USarXiv cs.CR (Security)

    arXiv:2605.19227v2 Announce Type: replace Abstract: Unified autoregressive models (UAMs) are transformer models that generate text as well as image tokens within a single autoregressive pass. Shared parameters and a multimodal vocabulary…

  105. NOMOS: Compiling Written Policies into Statically Verified Tool-Call Gates for LLM Agents
    USarXiv cs.CR (Security)

    arXiv:2610.11030v1 Announce Type: new Abstract: Tool-using LLM agents violate the policies they are deployed to enforce, often silently. Prior defenses hand-write rules, query an LLM verifier per action, or compile policies through…

  106. False Claims, Credible Images: A Red-Teaming Benchmark for Commercial Image Generators
    USarXiv cs.CR (Security)

    arXiv:2610.11112v1 Announce Type: new Abstract: Image-generation models can now produce text-rich, natural-looking visual artifacts that are hard to distinguish from real-world evidence, such as news reports and textbook pages. Yet, the…

  107. A Survey of Security Research for Operating Systems
    USarXiv cs.CR (Security)

    arXiv:2610.11471v1 Announce Type: new Abstract: In recent years, information systems have become the social infrastructure, so that their security must be improved urgently. In this paper, we introduce the results of the survey of…

  108. Where Do the Tokens Go? Understanding and Reducing Costs in LLM Agents for Vulnerability Discovery
    USarXiv cs.CR (Security)

    arXiv:2610.11602v1 Announce Type: new Abstract: LLM agents can spend millions of tokens during vulnerability discovery without producing a working proof of concept (PoC). What consumes that budget, and why does it fail to produce results?…

  109. Host Attack Graph for Botnet Propagation
    USarXiv cs.CR (Security)

    arXiv:2610.11640v1 Announce Type: new Abstract: Botnets represent the backbone of most modern cybersecurity attacks through which botmasters deploy distributed denial of service and advanced persistent threats against legitimate computer…

  110. The Hint Weight of ML-DSA Signatures Is Key-Dependent: An Empirical Study across the Three FIPS 204 Parameter Sets
    USarXiv cs.CR (Security)

    arXiv:2610.10992v1 Announce Type: new Abstract: Every ML-DSA (FIPS 204) signature carries a public hint vector h. We find that the Hamming weight of each hint polynomial h_k depends on the signing key: to first order it measures the…

  111. HPQ-AKE: A Provably Secure Sign-Less Hybrid Authenticated Key Exchange Protocol for Bandwidth-Constrained IoT and Edge Networks
    USarXiv cs.CR (Security)

    arXiv:2610.12024v1 Announce Type: new Abstract: Securing bandwidth-constrained Internet of Things (IoT) and edge networks during post-quantum migration creates an authentication trade-off: ML-KEM provides post-quantum key establishment…

  112. Protecting CPU AI On Edge TEEs: WebAssembly's Promise and Practical Challenges
    USarXiv cs.CR (Security)

    arXiv:2610.12050v1 Announce Type: new Abstract: AI models on edge hardware contain important intellectual property (IP), but an adversary can steal it when they achieve root access. Trusted Execution Environments (TEE) like Arm TrustZone…

  113. Poster: A Preliminary Study of LLM Distillation Inference
    USarXiv cs.CR (Security)

    arXiv:2610.12137v1 Announce Type: new Abstract: Unauthorized model distillation, in which a model is trained on the outputs of a proprietary large language model (LLM), is a growing threat to model providers. We study distillation…

  114. Secure Aggregate Encryption with Identity-Based Authentication for Multi-Vendor FPGA Cloud Deployment
    USarXiv cs.CR (Security)

    arXiv:2610.11873v1 Announce Type: new Abstract: Secure deployment of FPGA bitstreams in heterogeneous multi-vendor cloud infrastructures requires scalable authorization, authenticated device access, and bitstream protection throughout the…

  115. From Reactive Containment to Proactive Assurance: Lessons from OpenAI, Anthropic, and Google Agent Security Incidents
    USarXiv cs.CR (Security)

    arXiv:2610.12463v1 Announce Type: new Abstract: In 2026, cybersecurity evaluations involving OpenAI, Anthropic, and Google agents reached real systems outside their authorized test scope. The paths were different. OpenAI agents exploited…

  116. Moving Target Defense in SDN-enabled EV Charging Network
    USarXiv cs.CR (Security)

    arXiv:2610.11996v1 Announce Type: new Abstract: Software-Defined Networking (SDN) is emerging as a promis- ing technology for EV Charging infrastructure (EVCI) since it enables flexible, programmable control of EVCI networks, yet its…

  117. One Node, Two Roles: Simultaneous Contests for Validation and Attention in Rollups
    USarXiv cs.CR (Security)

    arXiv:2610.12220v1 Announce Type: new Abstract: An optimistic rollup is safe as long as at least one honest validator executes its state transition function (STF) and disputes any assertion that is inconsistent with the result of the…

  118. Skill Constellations: Tracing the Supply Chain of Agent Skills on GitHub
    USarXiv cs.CR (Security)

    arXiv:2610.11169v1 Announce Type: cross Abstract: Agent skills are SKILL.md instructions and scripts that AI coding agents such as Claude Code and Codex run with the permissions of their user. Developers share skills by copying them…

  119. Stop My Dancing! Understanding, Detecting and Attributing Motion-Aware Deepfake Videos
    USarXiv cs.CR (Security)

    arXiv:2610.11496v1 Announce Type: cross Abstract: Pose-guided diffusion models can now synthesize entire human figures in motion, spawning a new class of deepfakes: Motion Aware Deepfake (MAD) that have already reached hundreds of millions…

  120. Local Sensitivity in Exponential Selection: Failure Modes and Valid Calibrations
    USarXiv cs.CR (Security)

    arXiv:2610.11870v1 Announce Type: cross Abstract: Selection is a task that chooses one element from a finite public candidate range to maximize a data-dependent score. In differential privacy (DP), the exponential mechanism (EM) samples a…

  121. Phantom Transfer: Data Poisoning can Survive Data-Level Defences
    USarXiv cs.CR (Security)

    arXiv:2602.04899v3 Announce Type: replace Abstract: We present a data poisoning attack -- Phantom Transfer -- with the property that, even if you know precisely how the poison was placed into an otherwise benign dataset, you cannot filter…

  122. Rethinking Latency Denial-of-Service: Attacking the LLM Serving Framework, Not the Model
    USarXiv cs.CR (Security)

    arXiv:2602.07878v2 Announce Type: replace Abstract: LLM inference is inherently expensive, even a modest slowdown can translate into substantial operating costs and severe availability risks. Recently, a growing body of research known as…

  123. Context-Binding Gaps in Stateful Zero-Knowledge Proximity Proofs: Taxonomy, Separation, and Mitigation
    USarXiv cs.CR (Security)

    arXiv:2604.03900v2 Announce Type: replace Abstract: A zero-knowledge proximity proof certifies geometric nearness but carries no commitment to an application context. In stateful geo-content systems, where drops can share coordinates…

  124. Recorded Future Introduces AI Infrastructure Indicator Lists, Strengthening AI Governance
    USRecorded Future

    Today, Recorded Future is announcing AI Infrastructure Indicator Lists , curated datasets for security teams to identify, monitor, and implement controls for AI-related network traffic. AI Infrastructure Indicator Lists are included for…