Cyber security
8 items · 48 articles in this topic
Besides the OpenSSH, apt and Zyxel stories, the day was dominated by unpatched or unpatchable products and by disruptive attacks on public-sector and critical services. Several vendors are declining to fix flaws, leaving defenders to mitigate on their own.
- 1
Anthropic declined to fix a Claude Dispatch flaw that could let attackers escalate from low privileges to remote command execution across corporate boundaries. The Register reports the refusal, so affected organizations must mitigate themselves.
- 2
A cyberattack on a refrigeration equipment provider altered temperature and defrost settings at three French national blood service facilities, endangering stored blood. This is a physical-safety consequence of a supplier compromise.
- 3
Norway's aid agency Norad was breached and all user data was stolen. The agency cannot confirm whether sensitive records of people in conflict areas are safe.
- 4
Queens Public Library in New York is dealing with ransomware that took Wi-Fi, checkouts and its website offline. No group has claimed it yet.
- 5
SonicWall will not patch end-of-life Global Management System 9.5.2 and earlier. CVE-2026-109593 (CVSS 9.1), an unauthenticated file read, stays unfixed, and a second flaw can alter database records. Operators must migrate or isolate.
- 6
DreamHost says attackers exploited a vulnerability in a deliberate, months-long attack to access customer emails and files.
- 7
CISA added CVE-2024-27834, an Apple OS authentication flaw already patched, to its KEV catalog because it is being exploited to deliver info-stealing malware. Unpatched Apple devices remain the target.
- 8
A phishing campaign abused Google Calendar invites and fake login pages to hijack corporate Google sessions and bypass MFA. Nearly 900 domains were taken down.
More in Cyber security · 40 articles›
OpenSSH 10.6p2 fixes a critical flaw on musl libc Linux systems without PAM, such as Alpine, which admins should mitigate immediately.
LWN.netcveopensshalpineauthenticationFeed excerpt
Version 10.6p2 of OpenSSH has been released with a workaround for a critical vulnerability that affects Linux systems that use musl libc with PAM disabled, such as Alpine Linux. OpenSSH developer Damien Miller said: If you operate a Linux system that uses musl libc without PAM (Alpine Linux is one example) then stop reading and set PasswordAuthentication=no or disable SSH for these systems IMMEDIATELY. This vulnerability does not exist on non-Linux systems, Linux/glibc systems and Linux systems that use PAM for password authentication. We consider the root cause of this bug to be outside OpenSSH itself, but we're in a good position to deliver a mitigation quickly. The Alpine Linux project has announced that it will be backporting the OpenSSH fix "all the way back to Alpine 3.16", which was released in May 2022.
North Korean operatives used a fake job interview to breach Binance, resulting in a reported $6.3B cryptocurrency heist.
Hacker News (front page)north-koreacryptobreachsocial-engineeringFeed excerpt
Article URL: https://www.theregister.com/security/2026/11/03/north-korean-fake-job-interview-led-to-63b-binance-heist/5307398 Comments URL: https://news.ycombinator.com/item?id=50365755 Points: 86 # Comments: 45
OpenSSH 10.6p2 released a critical workaround for Alpine Linux builds that lack PAM support.
Hacker News (front page)cveopensshlinuxpatchFeed excerpt
Article URL: https://www.openssh.org/releasenotes.html#10.6p2 Comments URL: https://news.ycombinator.com/item?id=50377976 Points: 41 # Comments: 3
OpenAI reports its autonomous vulnerability discovery found 948 valid local privilege escalation bugs in the Linux kernel.
Phoronixlinuxopenaikernelprivilege-escalationFeed excerpt
OpenAI published a blog post outlining their work on autonomous vulnerability discovery. To showcase their AI potential, they found 948 valid local privilege escalation vulnerabilities within the Linux kernel...
OpenSSH 10.6p2 ships a workaround for a critical vulnerability, so admins should update or apply the mitigation.
Hacker News (front page)opensshpatchvulnerabilityFeed excerpt
Article URL: https://hachyderm.io/@damienmiller/117563026759098135 Comments URL: https://news.ycombinator.com/item?id=50377221 Points: 41 # Comments: 11
The GhostMap Android flaw lets a permission-free app force devices into a persistent boot loop once installed.
BleepingComputerandroidvulnerabilitymalwareFeed excerpt
Attackers can force Android devices into a continuous boot loop using a specially crafted app that requires no permissions and only needs to be installed. [...]
CERT-In warns that the AI-assisted threat actor Crafty Falcon is actively targeting Indian banking, financial and insurance organizations.
The Hacker Newscert-inaithreat-actorFeed excerpt
The Indian Computer Emergency Response Team (CERT-In) has disclosed that an artificial intelligence (AI)-assisted threat actor it tracks as Crafty Falcon is targeting organizations in the Banking, Financial Services, and Insurance (BFSI) sectors in the country. Describing the threat as active and ongoing, the activity involves the use of publicly-available AI-driven offensive security tools
Binarly found Zyxel GS1900 switches remain exploitable via a root command execution zero-day after the earlier patch; new firmware fixes 10 models.
The Hacker Newszyxelzero-dayrcefirmwareFeed excerpt
Zyxel GS1900 network switches that received the vendor's earlier fix for an actively exploited flaw are still open to the exploit attackers used, researchers at Binarly found. The exploit lets anyone who can reach a switch's web management interface run commands on it without logging in. Zyxel has released new firmware for 10 GS1900 models that closes the hole. The earlier fix is the 2.90
Krebs reports White House staff devices run residential proxy services, including one from a Russian firm tied to 2016 election attacks.
Krebs on Securityproxywhite-houserussiaFeed excerpt
Several staffers at the White House are using computers or mobile devices equipped with residential proxy services that render their internet address available for others to rent, security experts warned today. One of the proxy services is a Russian company whose founder's web hosting business was ground zero for cyberattacks against election systems in Arizona and Illinois in 2016.
OpenAI reports an AI pipeline validated and patched 948 Linux kernel local privilege escalation vulnerabilities for about $1.2 million.
The Hacker Newslinuxprivilege-escalationllmvulnerabilitiesFeed excerpt
OpenAI on Wednesday detailed a pipeline that employed an upcoming cyber-focused artificial intelligence (AI) model to validate and patch 948 local privilege escalation vulnerabilities in Linux for $1.2 million in API costs. "Our prototype identified and analyzed 2,308 candidate Linux kernel findings, dynamically reproduced 1,724 vulnerabilities, and produced 948 local privilege escalation (LPE)
OpenAI publishes research on local privilege escalation vulnerabilities in Linux kernel code paths.
Hacker News (front page)linuxprivilege-escalationcveopenaiFeed excerpt
Article URL: https://openai.com/index/a-thousand-fewer-paths-to-root/ Comments URL: https://news.ycombinator.com/item?id=50372929 Points: 13 # Comments: 6
Researchers report a group, possibly Kremlin-sponsored, exploited iOS and Android flaws to infect phones of diplomatic targets.
Ars Technicamobilespywarethreat-actoriosandroidFeed excerpt
Researchers say group may have been Kremlin-sponsored in search of diplomatic targets.
Zyxel patched a zero-day in its GS1900 switches that attackers exploited to compromise nearly 1,000 devices.
BleepingComputerzero-daycvenetwork-devicesFeed excerpt
Zyxel has patched a zero-day vulnerability in its GS1900 series network switches that hackers exploited to compromise nearly 1,000 devices worldwide and steal configuration files and credentials. [...]
An APT flaw lets man-in-the-middle attackers swap verified Debian and Ubuntu packages, running code as root during download retries.
The Hacker NewscveaptdebianubuntulinuxFeed excerpt
A flaw in APT, the package manager for Debian and Ubuntu, allows an attacker who can modify package downloads to swap a checked package for their own. The attacker's code then runs as root, according to advisories that Debian, Ubuntu, and APT's developers published on November 4. The attack needs APT to retry failed downloads, which it does by default from Debian 12 and
Hackers tampered with NIKKE game launcher scripts in a supply-chain attack, distributing credential and crypto-wallet stealing malware.
BleepingComputersupply-chainmalwaregamingFeed excerpt
The developers of Goddess of Victory: NIKKE have confirmed that hackers tampered with the game's PC launcher scripts in a supply-chain attack, distributing malware capable of stealing credentials and cryptocurrency wallet data. [...]
Check Point Research describes 'prompt adware' hidden in a Haiou jailbreak bundle that injects ads into OpenAI Codex configuration.
The Hacker Newsai-agentjailbreakadwareFeed excerpt
Cybersecurity researchers have disclosed details of an OpenAI Codex jailbreak bundle that's designed to modify an artificial intelligence (AI) agent's trusted instructions to insert advertisements for a third-party AI gateway. The phenomenon has been codenamed prompt adware by Check Point Research. Called Haiou (meaning Seagull in Chinese), the jailbreak modifies OpenAI Codex's configuration
Binance says hackers behind billions in crypto thefts spent months planning, using laptop infections and fake job interviews.
Ars Technicacryptoheistsocial-engineeringFeed excerpt
Techniques included laptop infections, compromised internal tooling, and fake job interviews.
Debian published a security advisory for the apt package manager addressing a man-in-the-middle vulnerability.
Hacker News (front page)debianaptadvisorymitmFeed excerpt
Article URL: https://lists.debian.org/debian-security-announce/2026/msg00526.html?pow_referer= Comments URL: https://news.ycombinator.com/item?id=50366575 Points: 7 # Comments: 0
SonicWall disclosed two security flaws in end-of-life Global Management System software that will not receive patches.
BleepingComputersonicwallunpatchedend-of-lifevulnerabilityFeed excerpt
SonicWall warned customers of two security flaws affecting end-of-life Global Management System (GMS) systems that will not be patched. [...]
APT29 is linked to a watering hole campaign using a compromised Plerdy SEO Tracker script to deliver iOS and Android exploits.
The Hacker Newsapt29watering-holemobilesupply-chainFeed excerpt
The Russia-linked threat actor known as APT29 is said to be behind a large-scale watering hole campaign targeting iOS and Android users through compromised websites, fake domains, and a supply chain attack involving a Ukrainian web analytics service called Plerdy SEO Tracker. "The injected JavaScript code collects browser and device attributes," Sekoia researchers Amaury-Jacques Garçon and
A paper shows that differences between Office document extractors and display can let adversaries mislead LLMs reading uploaded files.
arXiv cs.CR (Cryptography & Security)llmprompt-injectionofficeFeed excerpt
arXiv:2611.05806v1 Announce Type: new Abstract: Office documents are now routinely uploaded to large language models (LLMs) for processing. However, an LLM does not always see what a human sees, because it typically reads the text that an extractor produces from the document's underlying Office Open XML (OOXML) files. Depending on the implementation, extractors differ in which OOXML parts they read and how they interpret them, so their output can differ from what Office applications display. For example, by slightly shifting the positions of text boxes on a slide, an adversary can swap the order in which the model reads them while the slide looks unchanged. We show that this extraction divergence is an attack surface for LLM applications. In this work, we first measure what an adversary can exploit by analyzing the widely used extractors that LLM applications rely on. By combining static source code analysis with dynamic analysis, we identify 19 OOXML channels where extractors read content a human cannot see and 12 where they drop content a human can. We also find that widely used extractors ignore display formatting and reading order, so even visible content reaches the model in a…
PIMROP demonstrates return-oriented programming on the commercial UPMEM DPU processing-in-memory instruction set.
arXiv cs.CR (Cryptography & Security)ropprocessing-in-memoryresearchFeed excerpt
arXiv:2611.06036v1 Announce Type: new Abstract: Processing-in-Memory places small programmable cores on the memory device, where they execute beside the DRAM banks and beyond the control-flow defenses that protect host programs. The security of this in-memory code is an attack surface that remains largely unstudied. We present PIMROP, to our knowledge the first demonstrated code-reuse attack on a commercial Processing in-Memory ISA. We show that return-oriented programming is feasible on a real, commercially shipped PIM instruction set, the UPMEM DPU, whose DRAM Processing Units run C compiled by a dedicated LLVM toolchain. We build Galileo-for-DPU, a gadget scanner adapted to the DPU's fixed-width, 8-byte-aligned RISC encoding, and run it entirely on the UPMEM functional simulator in a reproducible Docker and VM environment. We find the runtime library that every kernel links already yields 517 non-trivial gadgets covering all five canonical ROP primitive classes, and all 13 PrIM benchmarks we scanned carry gadgets as well. We then demonstrate two verified end-to-end attacks, each reproducible on 10 of 10 runs. In the first, a stack buffer overflow in a DPU kernel overwrites the s…
SenTrojan presents an analog hardware Trojan framework that implants hidden attack surfaces into sensors before deployment.
arXiv cs.CR (Cryptography & Security)hardware-trojansensorsresearchFeed excerpt
arXiv:2611.06103v1 Announce Type: new Abstract: Sensors bridge the physical and digital worlds, making their integrity critical to cyber-physical system security. Existing sensor attacks largely exploit unintended and sensor-specific physical vulnerabilities, limiting their generalizability and motivating a different threat model, i.e., whether an adversary can implant hidden attack surfaces into sensors before deployment. In this paper, we present SenTrojan, an analog hardware Trojan framework for sensors. SenTrojan embeds malicious functionality into analog signal circuitry commonly shared across diverse sensor architectures, and activates the hidden behavior through crafted external physical signals. Once activated, SenTrojan can induce sensor malfunctions ranging from permanent denial-of-service to static or dynamic spoofing. To achieve these effects under different architectural constraints, we introduce two complementary designs. The Logic-Rule-Driven (LRD) Trojan adds lightweight analog monitoring logic to detect trigger features such as frequency and amplitude, enabling reliable activation across various sensor types. The Physical-Effect-Driven (PED) Trojan targets sensors …
Researchers used photon-emission microscopy and laser fault injection to bypass secure debug protection on the RP2350A4 microcontroller and recover a test secret.
arXiv cs.CR (Cryptography & Security)hardware-attackfault-injectionrp2350Feed excerpt
arXiv:2611.06386v1 Announce Type: new Abstract: Fault-resistant policy storage does not guarantee fault-resistant enforcement. We demonstrated this distinction on the RP2350A4 using differential photon-emission microscopy (PEM) and laser fault injection (LFI). PEM narrowed the area for laser scanning, which found two sites where pulses set the DEBUGEN bits enabling core~1 Secure debug. An online attack sequence using only Serial Wire Debug (SWD) feedback restored sustained access. After rescue reset, the sequence recovered a 128-bit test secret. This reset reloads the secret's runtime page lock from a stored policy that permits Secure reads and prevents user firmware from tightening the lock.
Paper shows signed offline RAG knowledge packs can have text-vector mismatches letting attackers steer retrieval, and proposes closing the gap.
arXiv cs.CR (Cryptography & Security)ragllm-securityintegrityFeed excerpt
arXiv:2611.06498v1 Announce Type: new Abstract: Offline retrieval-augmented generation (RAG) on edge devices is updated through signed knowledge packs that carry text and precomputed embedding vectors. A signature proves who published a pack, not that its vectors encode its text. We study this text-vector integrity gap. A verifier that recomputes vectors and accepts them within a cosine tolerance concedes a predictable share of queries to an attacker who controls the vectors: the hijack rate equals the distribution of angular retrieval margins at the tolerance (the margin law). Margins are small in five corpora and two embedding models, so an idealised tolerance at the level of floating-point noise leaves up to 5.5% of queries exposed (at most 1.2% in any corpus once exact duplicate documents are excluded), and a threshold calibrated from our CPU INT8-FP32 comparison of a small embedding model (at the empirical 0.1% quantile of honest cosines) leaves 98-100% exposed; on a Raspberry Pi 5 running a local language model, such an attack lowers correct answers from 0.75 to 0.39. Integrity audits of compressed vector stores fare no better: on a FAISS product-quantisation index over 57,63…
Researchers demonstrate a signed MCP tool that leaks sensitive data past Meta's MCPGuard-Dynamic runtime security policy.
arXiv cs.CR (Cryptography & Security)mcpai-agentsdata-leakFeed excerpt
arXiv:2611.06585v1 Announce Type: new Abstract: AI agents rely on third-party tools, such as Model Context Protocol (MCP) servers, whose publishers can be authenticated and whose execution can be restricted by runtime security policies. Yet authenticity does not guarantee benign behaviour: a legitimately signed tool may perform its advertised function while concealing malicious code that leaks confidential information. We demonstrate this trust failure with a signed MCP tool that embeds sensitive data in requests to a policy-allowed destination. Meta's MCPGuard-Dynamic, a state-of-the-art runtime defence, blocks a direct exfiltration attempt in the same run, yet fails to prevent this attack in every trial. We introduce germline execution, which limits information disclosure without inspecting or trusting tool implementations. Each tool invocation runs in ephemeral copies forked from a data-free germline process, each receiving only information authorised for its destination. Even if malicious code executes, what it can disclose is bounded by the data available to its copy. Under explicit assumptions on copy confinement and a data-free germline, we prove sink noninterference for eve…
Indirect Prompt Injection in the Wild: An Empirical Study of Prevalence, Techniques, and Objectives3
Large-scale empirical study measuring how often indirect prompt injections appear in real webpages and what objectives they pursue.
arXiv cs.CR (Cryptography & Security)prompt-injectionllmwebFeed excerpt
arXiv:2604.27202v2 Announce Type: replace Abstract: As LLMs are increasingly integrated into systems that browse, retrieve, summarize, and act on web content, webpages have become an untrusted input vector for downstream model behavior. This enables site owners, contributors, and adversaries to embed instructions directly in web resources, i.e., indirect prompt injections. While prior work demonstrates such attacks in controlled settings, their prevalence, deployment, and real-world impact remain unclear. We present one of the first large-scale empirical analyses of indirect prompt injections in webpages and HTTP responses. Analyzing 1.2B URLs from 24.8M hosts, we identify 15.3K validated instances across 11.7K pages. These are not isolated cases: a small number of recurring templates account for most cases. We characterize their objectives, delivery mechanisms, visibility, persistence, and impact, revealing a heterogeneous ecosystem spanning disruptive prompts, reputation manipulation, content-protection directives, and AI-bot detection, targeting systems such as crawlers, search pipelines, customer-support agents, and hiring workflows. A key finding is that most instructions targ…
Position paper arguing agent security should treat the AI model as untrusted and enforce invariants at the system level.
arXiv cs.CR (Cryptography & Security)agentsposition-paperFeed excerpt
arXiv:2605.18991v3 Announce Type: replace Abstract: We take the position that agent security must be approached as a systems problem: the AI model powering the agent must be treated as an untrusted component, and security invariants must be enforced at the system level. Through this lens, efforts to increase model robustness (the dominant viewpoint in the community) are insufficient on their own. Instead, we must complement existing efforts with techniques from the systems security domain. Based on our experience as cybersecurity researchers in operating systems, networks, formal methods, and adversarial machine learning, we articulate a set of core principles, grounded in decades of systems security research, that provide a foundation for designing agentic systems with predictable guarantees. As evidence, we analyze eleven representative real-world attacks on agents and discuss how systems principles, if realized, could have prevented these attacks. We also identify the research challenges that stand in the way of implementing these principles in agents.
CodeSpear shows grammar-constrained decoding, used to enforce syntactic validity, can be exploited to jailbreak LLMs into generating malicious code.
arXiv cs.CR (Cryptography & Security)jailbreakllmcode-generationFeed excerpt
arXiv:2606.11817v3 Announce Type: replace Abstract: Large Language Models (LLMs) are increasingly used for code generation, raising concerns that they may be misused to produce malicious code. Meanwhile, Grammar-Constrained Decoding (GCD) has been widely adopted to improve the reliability of LLM-generated code by enforcing syntactic validity. In this paper, we reveal a counterintuitive risk: this reliability-oriented technique can itself become an attack surface. We uncover a new jailbreak attack, termed CodeSpear, that exploits GCD to induce LLMs into generating malicious code. Our experiments show that simply applying a benign code grammar constraint can effectively jailbreak LLMs. To address this vulnerability, we propose CodeShield, a safety alignment approach that robustly preserves safe behavior even under attacker-controlled grammar constraints. CodeShield aligns the model in the code modality by teaching it to generate honeypot code under GCD. Such code is semantically harmless, so it does not implement the malicious request, and structurally diverse, so it is difficult to suppress through grammar tightening. At the same time, CodeShield still preserves natural-language ref…
arXiv paper demonstrates jailbreaking of function-calling LLMs by injecting simulated moderation traces into shared model context.
arXiv cs.CR (Cryptography & Security)jailbreakllmprompt-injectionFeed excerpt
arXiv:2607.00481v2 Announce Type: replace Abstract: Jailbreak attacks remain a critical threat to the safe deployment of large language models (LLMs). While prior work has primarily studied attacks and defenses at the prompt level, we show that this prompt-centric paradigm overlooks a structural vulnerability in stateful, function-calling environments. In such applications, developer-defined schemas, structured arguments, and untrusted tool outputs are interleaved into a single shared model context. This architecture expands the attack surface by blurring the boundary between trusted control logic and untrusted data, allowing adversarial intent to be distributed across a multi-turn execution path. We exploit this architectural flaw through SMT, a black-box attack framework based on Simulated Moderation Traces. Departing from purely prompt-based interactions, SMT constructs a multi-turn trajectory that simulates a legitimate moderation-auditing workflow. Within this trajectory, a fabricated moderation frame leverages red-team testing as a pretext to elicit harmful generations. The subsequent validation feedback treats safety refusals as execution failures, prompting refinements that…
GhostTac paper shows electromagnetic interference can manipulate robotic tactile sensors without physical contact.
arXiv cs.CR (Cryptography & Security)physical-securitysensorsresearchFeed excerpt
arXiv:2608.20817v4 Announce Type: replace Abstract: Tactile sensors are integral components of modern robotic systems, enabling robots to perceive and interact with the physical environment through tactile feedback. Despite their importance, the physical-layer security of tactile sensors has received little attention in prior work. In this paper, we present GhostTac, to the best of our knowledge, the first contactless attack that manipulates tactile sensing via electromagnetic interference (EMI). We identify that EMI exploits the nonlinear rectification and limited bandwidth amplification effects, allowing carefully crafted EMI signals to be converted into a persistent DC offset that bypasses on-board filtering and induces stable measurement deviations. Building on this mechanism, GhostTac enables fine-grained and controllable manipulation of sensor outputs by reshaping the spatial distribution and manipulating the magnitude at the targeted location. Such interference can induce unintended and harmful robot behaviors, such as causing a domestic robot to exert excessive force, resulting in physical damage or human injury. We evaluate GhostTac on 10 sensor modules and 2 dexterous han…
X.js presents a live benchmark of 1,738 real Node.js taint-style vulnerability advisories covering injection and prototype pollution.
arXiv cs.CR (Cryptography & Security)nodejsnpmbenchmarkvulnerabilitiesFeed excerpt
arXiv:2610.18585v3 Announce Type: replace Abstract: JavaScript vulnerabilities in npm packages pose a significant risk to the Node$.$js ecosystem, and a growing body of work aims to detect, exploit, and mitigate them. Studying and evaluating this work requires up-to-date collections of real vulnerabilities, yet the most commonly used datasets stop recording vulnerabilities in 2022. We present X$.$js, a live benchmark of real-world Node$.$js taint-style vulnerabilities covering code and command injection, prototype pollution, and path traversal. Built from 1,738 advisories collected through mid-2026, primarily from the GitHub Advisory Database, X$.$js pairs each vulnerability with precise source and sink locations, a validated executable exploit and its execution trace, and, when available, a patched version of the package. X$.$js supports long-term reproducibility and growth: dependencies are mirrored in a self-hosted registry to guard against package removal from the ecosystem, and dedicated infrastructure generates and validates exploits for newly reported vulnerabilities. The current release contains 1,159 validated exploits for 965 packages. Using X$.$js, we characterize how th…
Systematic evaluation tests five jailbreak attack families and four defenses across seven open-weight LLMs including Llama-3 and Qwen2.5.
arXiv cs.CR (Cryptography & Security)jailbreakllmevaluationFeed excerpt
arXiv:2610.39509v3 Announce Type: replace Abstract: Large language models (LLMs) deployed in safety-critical settings must resist adversarial inputs designed to elicit harmful outputs. We present a systematic empirical evaluation of five adversarial jailbreaking attack families-direct requests, roleplay-based jailbreaking, hypothetical framing, instruction override, and obfuscation-across seven open-weight LLMs served locally: Llama-3-8B, Llama-3.1-8B, Llama-3.2-3B, Mistral-7B, Qwen2.5-7B, Gemma2-9B, and Phi-4-14B. We further evaluate four defense mechanisms-keyword blacklisting, self-reminder prompting, paraphrase-based sanitization, and perplexity-based filtering against two high-ASR models (Mistral-7B and Llama-3.1-8B) on a 20-prompt subset. Across 250 prompts per model, obfuscation attacks had the highest mean attack success rate (ASR) of any family (61.4%, under a rule-based judge with known defects that has not been validated against human labels), were the highest or tied-highest family for every model and achieved nonzero success on all seven. We find that model size does not significantly predict adversarial robustness (Spearman \r{ho} = -0.63, p = 0.13, n = 7), and that t…
PromptMIA introduces a membership inference attack exploiting soft prompts in federated prompt-tuning of foundation models.
arXiv cs.CR (Cryptography & Security)membership-inferencefederated-learningprivacyFeed excerpt
arXiv:2601.06641v4 Announce Type: replace-cross Abstract: Membership inference attacks (MIAs) pose a serious privacy threat in federated learning (FL). While MIAs have been extensively studied in standard FL, the recent shift toward federated fine-tuning introduces new and largely unexplored attack surfaces. In this work, we show that federated prompt-tuning, which adapts pre-trained foundation models using lightweight input prefixes, exposes a novel and effective vector for membership inference. We propose PromptMIA, a membership inference attack tailored to federated prompt-tuning, in which a malicious server introduces adversarially crafted prompts and exploits their updates during collaborative training to determine whether a target data point belongs to a client's private dataset. We formalize this threat via a security game and demonstrate that PromptMIA achieves consistently high attack advantage across diverse benchmark datasets, substantially outperforming current SOTA federated MIAs. We also provide a theoretical lower bound on the attack advantage that explains the observed empirical behavior. Finally, we show that existing MIA defenses are often ineffective against Prom…
A 2015 essay surveys the history of backdoors in computing and their security implications, resurfacing on Lobsters.
LobstersbackdoorshistoryessayFeed excerpt
Comments
A technical write-up details further improvements to exploits targeting the Xbox 360 hypervisor.
Hacker News (front page)xbox360hypervisorexploitFeed excerpt
Article URL: https://icode4.coffee/?p=1746 Comments URL: https://news.ycombinator.com/item?id=50372164 Points: 40 # Comments: 2
Vercel is rolling out Device Bound Session Credentials to defend against information-stealing malware session cookie theft.
BleepingComputervercelinfostealersession-securityFeed excerpt
Vercel is gradually rolling out Device Bound Session Credentials (DBSC) on its platform as a key defense against information stealer malware. [...]
Layer3 Intel reports residential proxy activity detected on White House network infrastructure.
Hacker News (front page)proxiesnetworkthreat-activityFeed excerpt
Article URL: https://layer3intel.com/blog/residential-proxies-white-house-network Comments URL: https://news.ycombinator.com/item?id=50371246 Points: 30 # Comments: 2
mSeven Software urges mSecure password manager customers to reset passwords and 2FA after a cyberattack exposed an encrypted database.
BleepingComputerbreachpassword-managerFeed excerpt
mSeven Software is warning current and former mSecure password manager customers to change passwords, reset two-factor authentication, and remove passkeys following a cyberattack that reportedly exposed an encrypted customer database. [...]
StreamVessel describes a redirect URI approach letting users trust MCP clients without enabling phishing attacks.
Hacker News (front page)mcpphishingoauthFeed excerpt
Article URL: https://streamvessel.com/blog/mcp-trusted-redirect-uris Comments URL: https://news.ycombinator.com/item?id=50367004 Points: 22 # Comments: 2