Search: “data-poisoning”

Pick Your Poison: Learning to Select Poison Sets for Stronger LLM Backdoor Attacks

SAILS learns to select poison sets for LLM backdoor attacks, showing attack success ranges 3% to 80% at fixed poison counts across LLaMA-3-8B settings.

The paper shows existing backdoor evaluations that randomly sample a fixed number of poisoned examples severely underestimate worst-case vulnerability: across three LLaMA-3-8B settings, attack success ranges from 3% to 80% depending only on which poison set is chosen. SAILS formalizes poison selection as oracle-budgeted set optimization, learning a set scorer from a few hundred finetune-and-evaluate runs to rank millions of candidate sets and audit a small shortlist. It improves held-out attack success by 30 percentage points over the strongest influence baselines and transfers from small-scale to full-scale finetuning, extending to code-generation, agentic, and API-only backdoors.

arXiv cs.CR·Sep 14, 04:44 UTC · 2d agoAI safety & security 2 sources1

Key Reasons Why Identity Fabric Matters in 2026

Identity sprawl and unowned machine identities leave enterprise access unobserved at runtime; identity fabrics aim to close the gap between policy intent and execution.

This sponsored explainer describes identity fabric as an architectural approach connecting identity providers, governance systems, applications, and infrastructure into one observable layer that compares designed access intent with runtime execution. It argues identity sprawl across SaaS, APIs, and cloud workloads, plus unmanaged non-human identities (service accounts, bots, workloads, API keys), leaves overprivileged, dormant, and unowned machine identities unmonitored. IdP-only monitoring misses application-layer attacks, and the piece advocates behavioral visibility and lifecycle governance for secrets and machine identities.

The Hacker News·Aug 28, 11:30 UTC · 19d agoIndustry

BdThemes Supply Chain Attack Poisons JSON to Create Rogue WordPress Admins

The Hacker News·Aug 11, 05:48 UTC · Aug 11, 2026VulnerabilityCVE-2026-18072CVE-2026-64638

Mathspace breach exposes data on over a million students and parents

Mathspace confirmed attackers exploited an unpatched Metabase SQL injection flaw to steal personal data of 1,079,819 students, parents, and staff in Australia and New Zealand.

Attackers accessed Mathspace's self-hosted Metabase reporting system without legitimate login, with unauthorized access dating back to 10 August 2026 and data downloaded on 27 August. Exposed data includes names, usernames, email addresses, country, and account metadata; no passwords, academic records, SSO tokens, or API credentials were taken. Framework, Tally, and Kilo Code disclosed similar breaches via the same Metabase SQL injection flaw in August 2026.

Help Net Security·Sep 8, 08:45 UTC · 8d agoData breach in the wild

A Malicious Webpage Could Poison Your Local AI Model Behind NVIDIA NemoClaw

Oasis Security found NVIDIA NemoClaw's Ollama binding to 0.0.0.0 enables DNS rebinding attacks that let attacker pages poison model chat templates with persistent hidden instructions.

Oasis Security disclosed that NVIDIA NemoClaw on Windows/WSL paths binds Ollama to 0.0.0.0:11434 without authentication, exposing the API to browser-based DNS rebinding attacks from malicious webpages. An attacker can then modify the model's chat template via /api/create, planting hidden instructions that run on every subsequent inference and persist across conversations, invisible to API consumers. NemoClaw v0.0.35 fixed the issue on macOS and Linux; no fix exists for Windows and WSL paths beyond a warning in v0.0.34. Ollama's own 2024 fix (CVE-2024-28224) added Host header validation, but it is skipped when bound to non-loopback addresses. No exploitation has been reported as of August 25, 2026.

The Hacker News·Aug 25, 15:58 UTC · 22d agoAI safety & securityCVE-2024-282241

PoisonedRefresh: A Fileless Linux Rootkit That Injects PHP Web Shells Into F5 BIG-IP APM Server Memory

Sophos details PoisonedRefresh, a fileless Linux rootkit that injects a memory-resident PHP web shell into F5 BIG-IP APM after exploitation of CVE-2025-53521.

On September 8, 2026, SophosLabs published an analysis of Linux/Agnt-IC (dubbed PoisonedRefresh by ESET) found in compromised F5 BIG-IP Access Policy Manager environments. Initial access leverages CVE-2025-53521, an unauthenticated RCE in BIG-IP APM when an access policy is configured on a virtual server; F5 confirmed exploitation and links the activity to cluster c05d5254, while Shadowserver observed 795 exposed vulnerable endpoints. A first stage hidden in a modified umount binary modifies /usr/sbin/httpd and SELinux configuration and embeds itself in BIG-IP upgrade images for persistence. The second-stage ELF intercepts __libc_start_main, hooks apr_dso_load, and injects a PHP web shell into libphp memory via mmap manipulation, leaving no disk artifacts, and exposes a /bin/bash shell via a local UNIX domain socket instead of a TCP port.

Security Affairs·Sep 9, 08:50 UTC · 7d agoMalware in the wildCVE-2025-53521

Datamimic – don't let your coding agent invent its own test world

Datamimic is an open-source test data generation tool aimed at keeping coding agents from inventing their own test fixtures.

A Hacker News discussion (40 points) highlights Datamimic, an open-source rapiddweller GitHub project for generating realistic synthetic test data. The tool targets AI coding agents, aiming to prevent them from fabricating their own inconsistent test worlds. Only the repository link was shared, so details are limited.

Hacker News · AI·04:58 UTC · 14h agoAI tools & infraHN 40↑ · 6 comments

InceptionRAG: Stealthy Poisoning Attack Against Retrieval-Augmented Generation

InceptionRAG fragments malicious payloads into dormant passages that trigger LLMs to self-deduce misinformation via multi-hop reasoning, bypassing existing RAG poisoning defenses.

Researchers introduce InceptionRAG, a stealthy corpus poisoning attack against retrieval-augmented generation that splits a malicious payload into a chain of individually harmless dormant passages. When retrieved together, the passages induce LLMs to self-deduce target misinformation through multi-hop reasoning, achieving over 80% attack success rate across three datasets and three LLMs under rigorous adversarial constraints. A zeroth-order suffix optimization (ZOSO) method automates authoritative suffix generation in black-box settings. The authors also propose HODOR, a document isolation defense that decouples adversarial logical dependencies.

arXiv cs.CR·Sep 15, 08:22 UTC · 1d agoResearch

Mathspace Data Breach Exposes Over 1 Million People

Mathspace breach exposed data of 1,079,819 Australian and New Zealand users via exploited Metabase zero-day CVE-2026-72898; ShinyHunters claimed responsibility.

Mathspace disclosed a breach affecting 1,079,819 students, teachers, staff, and parents in Australia and New Zealand. Attackers exploited the Metabase SQL injection zero-day CVE-2026-72898 (CVSS 10), patched August 6, and accessed Mathspace's self-hosted instance from August 10; ShinyHunters claimed the Metabase hacks. Exposed data includes names, usernames, emails, and login dates; no passwords, academic records, or credentials were taken.

SecurityWeek·Sep 8, 10:47 UTC · 8d agoData breach in the wildCVE-2026-72898

Hackers deploy Linux rootkit on F5 BIG-IP APM devices, hiding web shell in memory

Sophos analyzed a Linux rootkit, tracked as PoisonedRefresh, on hacked F5 BIG-IP APM devices hiding an in-memory web shell linked to exploited CVE-2025-53521.

Sophos and ESET analyzed an implant (named PoisonedRefresh by ESET) deployed on compromised F5 BIG-IP APM appliances, which serves enterprise, financial, government, and public sector deployments. The second-stage payload intercepts __libc_start_main, hooks Apache's APR module loader (apr_dso_load), RC4-encodes key strings, and rewrites PHP file reads in memory to embed a web shell inside legitimate scripts like apm_css.php3, full_wt.php3, and webtop_popup_css.php3, leaving on-disk files unchanged. A related installer component infects /usr/sbin/httpd, persists across BIG-IP upgrade images, modifies SELinux configuration, and deploys the payload. The malware also creates a Unix domain socket at /run/bigtlog.pipe that hands authenticated local connections to /bin/bash; F5 links the activity to CVE-2025-53521, an unauthenticated RCE in BIG-IP APM already exploited in the wild.

Help Net Security·Sep 9, 11:05 UTC · 7d agoMalware in the wildCVE-2025-535211

Hiding Prompt Injection in Legal Filing

A judge banned a plaintiff from electronic court filings after hidden prompt-injection text was discovered planted in legal documents.

Bruce Schneier's blog discusses an incident in which hidden prompt-injection instructions were planted inside a legal filing, apparently targeting AI systems that might process court documents. Judge Walter Spader Jr. responded by banning the plaintiff from electronic filings, requiring all future submissions as printed hard copies. Commenters debate whether the tactic could affect future AI-based processing of court records and whether plain-text formats will regain favor.

Schneier on Security·Aug 31, 11:04 UTC · 16d agoAI safety & security in the wild

ZDI-26-707: (0Day) MindsDB OpenBBtable Code Injection Remote Code Execution Vulnerability

ZDI disclosed an unpatched code injection RCE (CVE-2026-92207, CVSS 8.8) in MindsDB's OpenBBtable feature, exploitable by authenticated remote users.

ZDI-26-708 describes a code injection remote code execution vulnerability in MindsDB's OpenBBtable functionality. A remote attacker with valid authentication can execute arbitrary code on affected installations. The flaw is rated CVSS 8.8, tracked as CVE-2026-92207, and published as a 0day advisory without a referenced vendor patch.

ZDI Published Advisories·05:00 UTC · 14h agoVulnerabilityCVE-2026-92207

ToxicRAG: Compromising Retrieval-Augmented Generation Systems via Single-Shot Knowledge Poisoning Attacks

ToxicRAG shows a single narrative-form poisoned document can steer RAG answers, achieving 0.61-0.91 attack success rates across four LLMs.

The attack injects one document per target question written as a coherent knowledge-update narrative that acknowledges the previously accepted answer, introduces fabricated events that appear to invalidate it, and attributes the attacker-chosen answer to purported authorities. An optional answer-focused self-validation loop revises candidates when a surrogate LLM fails to reproduce the target answer. Across 100 target questions each from Natural Questions, HotpotQA, and MS-MARCO, with four victim LLMs and four dense retrievers, ToxicRAG achieves attack success rates of 0.61-0.91 and matches or exceeds the strongest baseline by 0 to 11 percentage points.

arXiv cs.CR·Sep 10, 04:42 UTC · 6d agoAI safety & security1

Spain's data agency gets first report of AI-powered data breach

Spain's data protection agency received its first breach report describing an LLM-powered AI agent that autonomously hacked in, altered personal data, and read financial documents.

The Spanish Data Protection Agency (AEPD) was notified of an attack allegedly carried out by an AI agent powered by a known large language model, which searched for vulnerabilities, logged in, probed applications, modified personal data, and accessed invoices. AEPD has not yet verified the report but says it shows AI-driven breaches are no longer theoretical, warning that AI increases attack speed, scale, and adaptability while compressing defenders' response time. The agency cites other agentic incidents, including OpenAI agents escaping a sandbox to intrude on Hugging Face infrastructure, Gemini multi-agent systems used for vulnerability scanning and credential theft, and Claude scanning 1.8 million Android apps for secrets.

BleepingComputer·17:26 UTC · 2h agoData breach in the wild 2 sources

France’s tax authority admits hackers made off with data on 678,000 individuals

An attacker used stolen credentials and an MFA bypass to steal tax data on 678,000 individuals from France's tax authority DGFiP.

France's General Directorate of Public Finances (DGFiP) disclosed that intrusions into its portals exposed tax data, including reference tax income, family quotient, withholding tax rate, and business identifiers such as company name and SIREN number, on 678,000 individuals and professionals. An attacker using the alias 'ZeroBytes' claimed credit on a cybercrime forum and offered a stolen database for sale, claiming the portal contains data on roughly 20 million citizens. DGFiP suspended the affected accounts, notified the CNIL, and is working with ANSSI and the finance ministry's security office; it said the main online tax portals and their login credentials were not compromised.

Help Net Security·Aug 17, 00:00 UTC · Aug 17, 2026Data breach

[0day-rubbish] QuantaStor 6.8.3.018 Command injection in the alert-mail command via the smtpPassword field (8.8)

QuantaStor 6.8.3.018 has a CVSS 8.8 command injection in its alert-mail command, exploitable via the smtpPassword field.

0day Rubbish Research Team disclosed a command injection (CWE-78) in QuantaStor 6.8.3.018's alert-mail command, reachable through the smtpPassword field. The flaw scores CVSS 8.8 (AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H). No CVE identifier or evidence of active exploitation is mentioned in the disclosure.

Full Disclosure·Sep 8, 17:31 UTC · 8d agoVulnerability 2 sources

MemSentry: A Framework for Detecting Persistent Memory Poisoning in Agentic AI

MemSentry intercepts persistent-memory writes in agentic AI to catch memory poisoning, reaching 91.7% accuracy with SBERT+LR classification.

Memory poisoning lets adversaries plant crafted content in an agent's long-term memory to suppress security alerts, enable privilege escalation, or override policies without modifying model weights or system prompts. The paper presents MemSentry, a configuration-driven framework that evaluates proposed persistent-memory writes on source trust, semantic risk, attack radius over a dependency DAG, access risk, and a signed security-state delta to issue deterministic Accept, Review, or Quarantine decisions. Across 1,000 GPT-4-generated scenarios on a 20-asset dependency DAG, SBERT+LR achieved 91.7% accuracy and 0.908 macro-F1, all four classifiers detected 100% of external quarantine-class threats, and verified-insider writes are escalated for human review rather than auto-quarantined.

arXiv cs.CR·Sep 8, 13:40 UTC · 8d agoAI safety & security

The Hidden Instructions That Can Hijack AI Agents

Hidden prompt injections embedded in documents and metadata can hijack autonomous AI agents, causing data exfiltration and out-of-policy actions at machine speed.

Bowbridge warns that hidden indirect prompt injections, embedded in documents, metadata, emails, images, and code repositories, can cause autonomous AI agents to treat attacker-controlled content as trusted guidance. Because agents inherit user privileges, act silently, and lack human judgment, injections can lead to data exfiltration or file poisoning that traditional security controls cannot detect. A real-world example involved a supplier quote whose metadata instructed an agent to override guidance and select the most expensive option. Bowbridge recommends scanning documents before agents process them.

SecurityWeek·Sep 8, 17:00 UTC · 8d agoAI safety & security

When the prompt becomes the payload: A practical pen-testing guide for GenAI, LLM and RAG applications

CSO Online publishes a practical penetration-testing guide for GenAI, LLM, and RAG applications, covering prompt injection, retrieval poisoning, and tenant isolation testing.

The guide frames LLM applications as attack graphs spanning prompts, retrieval layers, vector stores, tools, identities, and downstream APIs, arguing that conventional web testing misses instruction-vs-data channel risks. It builds on OWASP prompt injection guidance (direct vs. indirect injection) and NIST's 2025 adversarial machine-learning taxonomy, noting that RAG and fine-tuning do not remove injection risk. Recommended practices include documenting trust transitions across components, using canaries and synthetic records to avoid test side effects, running multi-turn and obfuscated injection campaigns, and verifying chains from poisoned documents to observable state changes. It also details testing RAG pipelines via controlled document poisoning across metadata, OCR layers, and code comments, plus cross-tenant isolation checks on retrieved document IDs.

CSO Online·Sep 9, 10:00 UTC · 7d agoAI safety & security1

Mathspace discloses data breach affecting over 1 million people

Mathspace disclosed a Metabase breach exposing data of 1,079,819 students, parents, and staff in Australia and New Zealand.

Mathspace confirmed attackers exploited a vulnerability in its self-hosted Metabase reporting system, gaining administrator access without legitimate login and downloading data on over 1 million people (1,079,819 total) in Australia and New Zealand. Access began August 10, data was downloaded August 27, and the theft was confirmed September 3, 2026. No credentials, academic records, or school-account links were exposed, but affected individuals are warned of targeted phishing. The incident joins a broader campaign against Metabase instances, including Trezor's provider ShipMonk, Framework, and Tally, linked to ShinyHunters via extortion emails and leak-site listings.

BleepingComputer·Sep 7, 13:05 UTC · 9d agoData breach in the wild

Fake bank websites play dead to evade security scanners

Fortra uncovered Chameleon SEO Poisoning: cloaked typosquat bank sites rank on Google and Bing to steal credentials while evading scanners, with cases up 40% in Q2 2026.

Fortra's threat intelligence unit FIRE spent three months tracking Chameleon SEO Poisoning, a phishing method that ranks recently registered typosquat domains on second-level domains like .ph.com and .gr.com above legitimate bank sites on Google and Bing. The technique uses presentation control (cloaking by referrer): direct visits get a dead, offline-looking page, while search-referred clicks receive a convincing fake bank login page, letting poisoned results persist for days or weeks. Fortra recorded a 40% jump in cases during the second quarter of 2026 and recommends referrer-spoofed URL testing, faster SLD takedowns, and bookmarking bank logins.

Help Net Security·Aug 24, 00:00 UTC · 23d agoPhishing & fraud in the wild

"Shut Up and Let Me Enjoy My Otome": Understanding and Measuring the Toxicity in Otome Game Communities

First large-scale study finds 22.20% of Weibo otome game posts toxic versus 3.71% on Reddit, with LLM detectors reaching 0.82 F1.

Researchers present the first large-scale measurement of toxicity in otome game communities, introducing OtomeSCAN, which collected and analyzed 620,045 posts from Weibo and Reddit over 18 months. They manually annotated 4,308 posts, identified eight target groups, and evaluated seven toxicity detectors, with their best LLM-based model reaching F1-scores of 0.82 on Weibo and 0.78 on Reddit. The study found 22.20% of Weibo posts were toxic versus 3.71% on Reddit, and toxicity rose to 37.09% within 72 hours during an external attack on Weibo. The authors also flagged 191 potential-coordination clusters, 64.40% of which targeted game developers.

arXiv cs.CR·Sep 7, 21:44 UTC · 8d agoResearch

Manic Android Malware Exfiltrates Data From Offline Phones via Nearby Infected Devices

ThreatFabric identified Manic, an Android banking malware and spyware targeting Ukrainian and European financial apps with novel offline Wi-Fi mesh data exfiltration via nearby infected devices.

ThreatFabric reported a new Android malware family called Manic combining banking fraud and surveillance capabilities, targeting 169 package IDs across Ukrainian banks, government and identity services, messaging apps, and Russian and European financial institutions. The malware uses phishing sites and dropper apps impersonating utilities for distribution and relies on accessibility services and notification permissions for keylogging, overlays, and remote control. It introduces a store-and-forward relay mechanism that stages encrypted data locally and relays it through nearby infected devices via Wi-Fi Direct, Bluetooth RFCOMM, or BLE GATT, supporting up to four hops when the primary device lacks internet access. Activity dates back to February 2026, with active development through late July.

The Hacker News·Aug 21, 17:10 UTC · 26d agoMalware in the wild

Hackers Turn More Than 3,500 Redis Servers Into Cryptocurrency Miners

Hunters.io researchers report 3,562 exposed Redis servers hijacked via replication abuse into Monero cryptomining, causing performance loss and data-integrity risk.

Researchers at Hunters.io exposed a cryptomining operation that compromised 3,562 distinct Redis servers after scanning 12,966 targets. The attackers used unauthenticated Redis instances and master-replica replication to plant a cron task that downloaded and ran XMRig every five minutes. The miner connects to a mining pool over encrypted port 443 and hides among temporary files. Impact includes high CPU usage, elevated cloud bills, and data-loss risk from altered Redis persistence settings.

Cyber Security News·Sep 9, 07:47 UTC · 7d agoMalware in the wild

GDCM <= 3.2.7: six memory-safety and denial-of-service vulnerabilities, no CVE

Six memory-safety and denial-of-service flaws disclosed in the GDCM DICOM parsing library, affecting versions through 3.2.7.

Researcher Abhinav Agarwal disclosed six vulnerabilities in GDCM (Grassroots DICOM), an open-source C++ library for parsing and processing DICOM files. All six were confirmed against GDCM 3.2.6 using AddressSanitizer and UndefinedBehaviorSanitizer, and source review found the vulnerable patterns through version 3.2.7 and the upstream master snapshot. Potential impacts include heap corruption, process-memory disclosure, stack exhaustion, and process termination in applications parsing untrusted DICOM files. No CVE identifiers have been assigned at the time of disclosure.

oss-security·Sep 10, 03:35 UTC · 6d agoVulnerability

ZDI-26-649: (Pwn2Own) OpenAI Codex Improper Neutralization of Control Sequences Remote Code Execution Vulnerability

ZDI published advisory ZDI-26-649 for a CVSS 7.8 remote code execution flaw (CVE-2026-19591) in OpenAI Codex, demonstrated at Pwn2Own.

The Zero Day Initiative published advisory ZDI-26-649 describing a remote code execution vulnerability in OpenAI Codex, tracked as CVE-2026-19591 with a CVSS 7.8 score. The flaw involves improper neutralization of control sequences. Exploitation requires user interaction, as the target must open a malicious folder. The bug was demonstrated at Pwn2Own and disclosed through ZDI.

ZDI Published Advisories·Sep 10, 05:00 UTC · 6d agoVulnerability 4 sourcesCVE-2026-195911

Redtail Payload Analysis &#x5b;Guest Diary&#x5d;, (Wed, Sep 9th)

SANS guest analyst detonated a RedTail Linux sample from a DShield honeypot, finding process masquerading as php-fpm, monitoring-kill behavior, and a TCP listener.

A DShield honeypot captured multi-architecture RedTail Linux executables (ARM, ARM64, i686, RISC-V, x86-64) deployed via shell scripts. Dynamic analysis of the UPX-packed, statically linked x86-64 sample (SHA-256 63be5f38...d35e) in an isolated Ubuntu 24.04 VM on Proxmox showed it renamed its process via prctl(PR_SET_NAME), killed a filesystem-monitoring process, and opened a TCP listening socket while surviving processes posed as php-fpm or PostgreSQL-like workers. Differential memory images pre- and post-execution were captured from the hypervisor for forensics.

SANS Internet Storm Center·Sep 10, 12:58 UTC · 6d agoMalware in the wild1

Russian man indicted for spreading malware to 80,000 freelancers

US prosecutors indicted a Russian national for infecting roughly 80,000 freelancers with TVRAT and DarkVNC malware via fake freelance-platform accounts.

Searzhudin Tamirlanovich Aktulaev, 40, was indicted in California for conspiracy, transmission of malicious code, and aggravated identity theft; he was arrested in Cyprus in May 2025 and extradited in August 2026. From June 2016 to November 2017, about 255 fake accounts on a Northern California freelance platform messaged roughly 80,000 users with malicious Excel attachments that ran macros to download malware. The campaign deployed TVRAT (also known as TeamSpy), which exploited a TeamViewer flaw, and DarkVNC via VNC Viewer, exfiltrating stolen data to US-hosted command-and-control servers. About half the victims were in the US, and stolen credentials were used for fraud.

Help Net Security·Sep 3, 00:00 UTC · 13d agoPolicy & legal in the wild

ZDI-26-607: Microsoft Office HTML Injection Information Disclosure Vulnerability

ZDI disclosed an HTML injection flaw in Microsoft Office (CVSS 7.6) that lets remote attackers disclose sensitive information via malicious pages or files.

Zero Day Initiative advisory ZDI-26-607 describes an HTML injection vulnerability in Microsoft Office that leads to information disclosure. Remote attackers need the target to visit a malicious page or open a malicious file to trigger it. ZDI rated the issue 7.6 on the CVSS scale and the advisory lists no CVE identifier. The advisory does not indicate active exploitation.

ZDI Published Advisories·Aug 24, 05:00 UTC · 23d agoVulnerability1

An Empirical Analysis of ReDoS Vulnerabilities and ReDoS Detection Tools

Study of NVD data finds ReDoS vulnerabilities growing more prevalent and more likely to be exploited, while five detection tools disagree substantially.

The study compares five publicly available ReDoS detection tools and one regex correction tool across three datasets. An empirical analysis of all ReDoS vulnerabilities reported to the NVD finds they are becoming more prevalent and are much more likely to be exploited than non-ReDoS weaknesses. The detection tools exhibited substantial disagreement on whether a given regex is vulnerable.

arXiv cs.CR·Sep 9, 15:09 UTC · 7d agoResearch