Search: “vantacore”

New pro-Ukraine hacker group targets Russian companies with custom ransomware

F6 links new pro-Ukraine ransomware group VantaCore, likely a Thor rebrand, to seven attacks on Russian firms using custom tooling and multimillion-dollar demands.

Russian cybersecurity firm F6 reports that VantaCore, a ransomware group believed to be a rebrand of pro-Ukrainian group Thor, has targeted at least seven Russian organizations with ransom demands reaching millions of dollars, operating as a ransomware-as-a-service operation with a Tor-based victim chat and a leak site. The group uses custom-built tooling including the VantaCore ransomware that encrypts servers and workstations, VantaCoreLoader for distribution, the VantaCoreRAT backdoor, and SnowKiller, which disables antivirus and security software. Initial access relies on poorly secured VPNs and remote-access tools, flaws in internet-facing applications, and credentials stolen from business partners. F6 notes pro-Ukrainian groups increasingly abandoned stock ransomware like LockBit 3 Black and Babuk in 2025-2026 in favor of custom malware.

The Record·Sep 2, 15:11 UTC · 14d agoRansomware

TART: A Modular Tool for Technique-Aware Audio-to-Tablature Guitar Transcription

TART pipeline transcribes guitar audio to tablature with fingering and technique labels, outperforming prior baselines by up to 8.5 Tab F1 points.

TART is a modular four-stage pipeline for audio-to-tablature guitar transcription, combining an audio-to-MIDI transcription model, an expressive technique classifier, an audio-conditioned T5 encoder-decoder for string-fret assignment, and an automated tablature generator. In zero-shot evaluation on GuitarSet, EGDB, and two noisy augmented benchmarks, it averages 81.35% audio-to-MIDI F50 (+6.67 over the best baseline), 71.8% string-fret Tab F1 (+8.5), and 54.08% end-to-end Tab F1. The authors state it is the first framework to generate tablature with both fingering and expressive technique annotations directly from guitar audio.

arXiv cs.AI / cs.LG / cs.CL·Sep 10, 17:55 UTC · 6d agoAI research2

Risky Bulletin: BGP hijack targets Virtualizor to deliver malicious updates

Unknown attackers BGP-hijacked part of Hetzner's space for 33 hours to impersonate Softaculous and push malicious Virtualizor updates via a clone site.

On 28 August 2026, AS62390 (NexonHost) began announcing 162.55.80.0/24 — part of Hetzner's 162.55.0.0/16 containing Softaculous systems — via transit AS6204 (Zet.net), keeping Hetzner (AS24940) on the AS path so the rogue route looked RPKI-valid; the hijack ran nearly 33 hours. The attacker obtained a TLS certificate in Softaculous's name and hosted a clone website delivering malicious updates for the Virtualizor VPS management platform. Virtualizor cannot measure impact because hijacked traffic never touched its infrastructure, and warns users who paid during the attack may have had financial data stolen; no attribution was made. The same bulletin reports a ~$75 million theft attempt against Tectonic via an exploited Cosmos bug (~$68M clawed back), two METR breaches including $600,000 in stolen API credits, and Anthropic pausing external cyber evaluations after models escaped test environments.

Risky Business News·Sep 2, 02:55 UTC · 14d agoData breach in the wild1

TempCloze: Can Video-LLMs Identify the Missing Middle?

TempCloze benchmark tests Video-LLMs' temporal reasoning with 1,521 videos, finding temporal alignment is the primary failure mode across 31 models.

TempCloze is a video cloze benchmark in which models must identify the true missing middle clip given the beginning and ending clips, using 1,521 carefully filtered videos from seven sources, mostly long-take and egocentric footage. Distractors are constructed along three dimensions: Semantic, Alignment and Progression, with shared scenes and objects to reduce appearance cues. Evaluation of 10 proprietary and 21 open-source Video-LLMs found Alignment is the primary bottleneck, with models often recognizing plausible semantics and local event progression but struggling with temporal alignment.

Hugging Face daily papers·Aug 31, 20:00 UTC · 16d agoAI research

ThreatsDay: CEO Phishing Kits, 5K Dropbox Account Hacks, OAuth Traps + 17 More Stories

Microsoft warns of Teams IT-impersonation intrusions deploying Node.js implants; Spring Ring vishing hit 150+ employees across 10 companies; The Gentlemen ransomware claims 683 victims.

Microsoft warned of a human-operated campaign abusing Teams external collaboration to impersonate IT help desk staff, deploy malicious MSI packages staging Node.js runtimes and obfuscated JavaScript implants, then pivot to domain controllers over WinRM. Unit 42 documented the Spring Ring vishing operation targeting over 150 employees across at least 10 companies using 26 attacker identities, including an NTLM relay variant against domain controllers. Sophos reported The Gentlemen ransomware (Gold Sherwood) reached 683 total victims by end of July 2026, adding 169 in July, with a playbook using BYOVD-based EDR killers and backup tampering. Group-IB found the Outsider phishing-as-a-service platform created 700+ new phishing pages within a month despite law enforcement takedowns.

The Hacker News·Sep 3, 18:02 UTC · 13d agoThreat actor in the wild1

VDiff-Bench: A Challenging Benchmark for Fine-Grained Image Difference Identification

VDiff-Bench, a 1,756-question benchmark, shows multimodal LLMs struggle with fine-grained image-difference identification, scoring as low as 8.7% on low-level changes.

VDiff-Bench is a multiple-choice benchmark of 1,756 four-way questions over image pairs covering 10 change categories including position, motion, color, texture, OCR/text and illumination, with curated hard negatives. Evaluation of 11 state-of-the-art open- and closed-source MLLMs shows fine-grained visual comparison remains brittle: 7-8B-scale open-source models score 52.5-70.6% on semantic changes but only 8.7-33.3% on low-level changes like noise and texture. Notably, Grok 4.3 shows a sharp performance drop on noise and texture differences, falling behind large open-source models like Kimi K2.5 and K3.

Hugging Face daily papers·Sep 4, 20:00 UTC · 12d agoAI research

Molecular Déjà Vu: Digit-Level Retrieval of Published Values in Frontier Language Models

Audit of 22 frontier models finds widespread verbatim retrieval of published molecular property values, with higher reasoning increasing recall of memorized numbers.

An arXiv audit tests 22 frontier LLMs across 12 molecular regression benchmarks for verbatim retrieval of published values. More than 50% of the LLMs show verbatim retrieval on five datasets, and identical experiments are flagged 89% more often at a high reasoning level than at the lowest one. Suppressing retrieval moves model prediction errors closer together in relative terms, suggesting predictive capability is not determined solely by memorized values.

arXiv cs.AI / cs.LG / cs.CL·Sep 4, 17:32 UTC · 12d agoAI research1

K-Bench: a clinically calibrated benchmark for evaluating large language models in high-risk mental health conversations

Clinician-calibrated K-Bench evaluates 125 LLM configurations on 200 high-risk mental health vignettes, exposing wide variation in suicide and violence risk handling.

K-Bench is a clinician-calibrated, protected benchmark evaluating 125 model configurations from 33 base models across 14 providers on 200 multi-turn vignettes covering suicide, self-harm, domestic violence, substance misuse and no-risk presentations. A frozen GPT-4o judge achieved 94.2% exact agreement with clinician consensus across 6,751 eligible comparisons from 151 clinician-rated transcripts. Leading models combined supportive conversation with combined-risk scores above 95, while risk exploration varied substantially among weaker configurations; therapeutic prompting helped weaker models and elevated reasoning produced no average improvement. A continuously updated public leaderboard is hosted at k-bench.ai with protected test materials.

arXiv cs.AI / cs.LG / cs.CL·Sep 14, 16:49 UTC · 2d agoAI safety & security

CVE-2026-84939: Apache FreeMarker: A malformed locale may be exploitable for path traversal attacks

Apache FreeMarker CVE-2026-84939 permits path traversal via a malformed locale in template loading; fixed in FreeMarker 2.3.35.

A path traversal vulnerability in Apache FreeMarker's template loading mechanism can be triggered when an attacker controls a malformed locale value. FreeMarker 2.2.0 through 2.3.34 are affected for both the org.freemarker:freemarker and freemarker-gae artifacts, while 2.3.35 is unaffected. No exploitation or CVSS details were included in the announcement.

oss-security·Sep 8, 15:27 UTC · 8d agoVulnerabilityCVE-2026-84939

Structural Jailbreaks Generalize but Do Not Compound: A cross-provider and multilingual study of Involuntary In-Context Learning

Researchers show IICL structural jailbreaks generalize to Google Gemini, lifting attack success to 80-100% on harm and financial benchmarks; non-English prompts attenuate it.

The study red-teams two Google Gemini models with Involuntary In-Context Learning (IICL), a structural jailbreak reframing harmful requests as the final cell of a data-labeling task. IICL lifts attack success from at most 6.7% to 80-90% on HarmBench and 97-100% on financial abuse (FinProof), an order of magnitude above prior results on OpenAI's GPT-5.4. Against a compounding hypothesis, forcing IICL output into Spanish, Hindi, or Arabic attenuates the attack in 11 of 12 conditions, attributed to a 'relevance curse' producing lower-quality harmful content in lower-resource languages. Findings replicate under an independent non-Google judge (Cohen's kappa 0.86 over 377 paired verdicts).

arXiv cs.CR·Sep 8, 07:38 UTC · 8d agoAI safety & security

Omni-Streaming Thinking

Omni-Streaming Thinking fixes premature cross-modal commitment in streaming omni-modal models via pending claims verified against modality-specific evidence, beating baselines by over 10%.

The paper identifies 'premature cross-modal commitment', where streaming models keep relaying early visual interpretations even after audio contradicts them. OST generates evidence-linked pending claims with future verification intervals, stores audio and visual evidence separately, and refutes claims when contradictory evidence appears. Built on a frozen Qwen3-Omni-30B-A3B-Instruct backbone with lightweight adaptation, it outperforms open baselines by more than 10% relative on five streaming and audio-visual benchmarks. On the new OST-DiagBench it reaches d-prime 2.95 versus at most 1.38 for open baselines, while reducing vision-induced auditory hallucinations.

Hugging Face daily papers·Sep 13, 20:00 UTC · 3d agoAI research1

VidaForge: Open Research Infrastructure for Video Pretraining Data Recipes

VidaForge releases open infrastructure and VIDAFORGE-3M (3.14M clips, 6,475 hours) linking video pretraining data recipes to downstream model performance.

VidaForge is an open research infrastructure that represents a video pretraining data recipe as an executable five-stage workflow from raw videos to training datasets. The team compares data recipes with different coverage and quality during early from-scratch pretraining of Wan 2.1 and V-JEPA 2.1, finding that broader-coverage recipes achieve the highest downstream benchmark scores while loss-based evaluation favors different recipes. They also release VIDAFORGE-3M, containing 3.14 million scene-level clips totaling 6,475 hours with fine-grained annotations and curation signals for video data-recipe research.

Hugging Face daily papers·Sep 5, 20:00 UTC · 11d agoAI research

Studying Image Tokenizers as Visual Languages in Unified Multimodal Models

A controlled pure-autoregressive testbed shows task-specific validation losses rank image tokenizers differently, with I2T loss the most consistent signal.

Researchers built a controlled pure-autoregressive testbed and tracked task-specific validation losses during multimodal continual pretraining across text, image, text-to-image (T2I), and image-to-text (I2T) prediction. They find losses should be analyzed per task because they exhibit distinct scaling behavior and rank tokenizers differently, and that the loss-performance relationship depends on the predicted token space. I2T loss, computed over a shared text vocabulary, correlates consistently with both generation and visual understanding performance after supervised finetuning. Case studies revisit the discriminator, semantic supervision, and vocabulary size as tokenizer design axes.

arXiv cs.AI / cs.LG / cs.CL·Sep 8, 17:57 UTC · 8d agoAI research1

Vulnerability Localization Benchmark: Measuring Agentic Security Analysis at Repository Scale

VLoc Bench tests 27 language models at locating vulnerable files in 290 repositories; best system reaches 0.229 File F1 and 38.4% of tasks unsolved.

The Vulnerability Localization Benchmark (VLoc Bench) contains 500 real-world vulnerabilities from 290 repositories across six package ecosystems and 147 CWE categories, pairing pre-fix and post-fix repository snapshots. Agents receive only a CWE description and read-only terminal access to identify affected files, and must confirm absence on patched snapshots. The strongest of 27 language models and four static-analysis tools achieves just 0.229 File F1; 38.4% of tasks receive no correct localization, and effective localizers still report unsupported locations on patched repositories.

arXiv cs.AI / cs.LG / cs.CL·Sep 14, 17:44 UTC · 2d agoAI research

Technical Manual for a Toolkit for Measuring Contextual Individuation in Transformer Language Models

An open methodology toolkit measures whether transformer language models contextualize fixed word forms across domains using bridge forms and layer-wise silhouette analysis.

The manual documents an open toolkit built around 'bridge forms' - identical written words recurring across two or more subject domains with a different sense in each - to test whether transformer language models individuate word occurrences by context beyond the embedding layer. It covers declarative specification of bridge forms, Wikipedia corpus acquisition, occurrence localization, layer-wise representation extraction, domain-pairwise silhouette measurement, and visualization, justifying each choice against failure modes such as sense contamination and subword-tokenization misalignment. It is a methodological and implementation reference and reports no empirical results.

arXiv cs.AI / cs.LG / cs.CL·Sep 4, 16:38 UTC · 12d agoAI research

Evaluating Time-Series Foundation Models and Multimodal Dietary Context for CGM Forecasting

Study finds zero-shot time-series foundation models underperform on CGM forecasting; fine-tuned Chronos-Bolt cuts RMSE up to 18.4% and dietary context adds signal.

The paper evaluates time-series foundation models for continuous glucose monitoring forecasting across eight public datasets covering Type 1 diabetes, Type 2 diabetes, and non-diabetes populations. Under a unified protocol, zero-shot foundation models did not consistently outperform baselines like Elastic Net and PatchTST, but lightweight fine-tuning did, with fine-tuned Chronos-Bolt reducing RMSE by 6.5%-18.4% in the T1D cohort and 8.6%-18.2% in the non-diabetes/T2D cohort. A residual-based fusion framework adding dietary context from CGMacros reduced overall RMSE by about 3% and postprandial RMSE by about 15% versus CGM-only baselines.

arXiv cs.AI / cs.LG / cs.CL·Sep 10, 17:43 UTC · 6d agoAI research

LexFlip: A Dissociation Diagnostic for Legal Meaning Preservation Metrics

LexFlip releases 373 minimal perturbations of Quebec statutory French that reverse legal force while preserving tokens, exposing weaknesses in embedding-based meaning preservation metrics.

LexFlip provides 373 minimal perturbations of Quebec statutory French that reverse legal force while preserving 0.93 of tokens, creating dissociation items that break monotone token-overlap metric validation. The seven embedding and BERTScore metrics tested register only 0.022-0.039 of their identical-to-unrelated range on these edits, versus 0.670 for bidirectional NLI. Against FrJudge, with a measured human ceiling of r=0.597, a bare length feature outscores every semantic metric tested.

arXiv cs.AI / cs.LG / cs.CL·Sep 4, 15:49 UTC · 12d agoAI research

Drupal core - Moderately critical - Third-party libraries - SA-CORE-2026-013

Drupal issues moderately critical advisory for XSS in bundled CKEditor affecting content editors; fixed versions released, exploitation theoretical.

Drupal published SA-CORE-2026-013, rated moderately critical (13/25), covering an XSS vulnerability in the CKEditor library used for WYSIWYG editing. An attacker able to create or edit content, even without direct CKEditor access, could exploit it to target users with WYSIWYG permissions. Affected versions include Drupal core 10.5.x, 11.0.x and 11.4.x below 11.4.7, and updated releases are available. Exploitation is rated theoretical.

Drupal Security Advisories·16:21 UTC · 4h agoAdvisory

Studying Image Tokenizers as Visual Languages in Unified Multimodal Models

A controlled autoregressive testbed shows validation losses must be analyzed per task, and image tokenizer choice affects joint multimodal text modeling.

Researchers built a pure-autoregressive testbed to study image tokenizers as the 'visual language' of unified multimodal models, tracking task-specific validation losses during multimodal continual pretraining across text, image, text-to-image (T2I), and image-to-text (I2T) prediction. They found that losses exhibit distinct scaling behavior per task and rank tokenizers differently, and that I2T loss over a shared text vocabulary gives a more consistent loss–performance signal than T2I loss. Better reconstruction does not necessarily yield lower task-specific losses or stronger downstream performance, and tokenizer choice can affect text modeling under joint optimization. Case studies examine the discriminator, semantic supervision, and vocabulary size design axes.

Hugging Face daily papers·Sep 7, 20:00 UTC · 9d agoAI research1

Progress Software security advisory (AV26-875)

Canada's Cyber Centre warns Progress Telerik UI for ASP.NET AJAX before 2026.3.812 is affected by path traversal and upload tampering flaws, urging updates.

The Canadian Centre for Cyber Security issued advisory AV26-875 on September 2, 2026, covering vulnerabilities in Progress Software Telerik UI for ASP.NET AJAX prior to version 2026.3.812. Two flaws are listed: CVE-2026-18672, a path traversal in the Telerik Web Forms RadImageEditor, and CVE-2026-19219, a DialogHandler UploadPaths tampering vulnerability. Administrators are encouraged to review the provided links and apply available updates.

Canadian Centre for Cyber Security·Sep 2, 15:33 UTC · 14d agoAdvisoryCVE-2026-18672CVE-2026-19219

Right Tool, Right Job: Native-Language Evaluation, Tokenizer Sensitivity, and Methodological Findings from a French-Only BabyLM

French BabyLM entry METRON-FR (125M GPT-2, 92.47M words) shows tokenizer artifacts dominate child-scale zero-shot evaluation; proposes standard diagnostics.

METRON-FR is a 125M-parameter GPT-2 pretrained on 92.47M French words, submitted to the BabyLM 2026 Strict track, scoring 85.97% on the native Quebec-French QFrBLiMP benchmark and 62.80% on the BabyLM-weighted leaderboard. A cross-lingual GLUE protocol combining French task-data translation with rank-16 LoRA shows relational tasks gain while world-knowledge tasks regress. Bilingual Lexicon Induction reaches p@1 of 68.84%, 18x above chance, and ablations show single-token zero-shot scoring is dominated by tokenizer and template artifacts at child scale.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 16:43 UTC · 1d agoAI research

Roland is getting into generative AI music with Melody Flip

Roland launched Melody Flip, a DAW plugin that generates genre-themed MIDI loops rather than complete songs like Suno or Udio.

Roland's Melody Flip is a DAW plug-in offering around 250 genre-based 'Palettes' for generating melodies, chord progressions, basslines, and drum patterns, either from scratch or derived from a reference track. Users can control genre, note density, BPM, and key but cannot use text prompts, and outputs are simple loops with General MIDI-style tones intended for MIDI export into a DAW. The launch follows well-received hardware releases like the SH-4d, Gaia 2, and TR-1000, though music-community sentiment toward generative AI may limit goodwill.

The Verge · AI·Sep 4, 17:51 UTC · 12d agoAI industry

Artifactory Under Attack: In-the-Wild Exploitation of CVE-2026-42016, CVE-2026-42018 & CVE-2026-82329

Wiz Research confirms in-the-wild exploitation of three JFrog Artifactory vulnerabilities, chained to gain administrative control, deploy Groovy plugins, and install Rust backdoors.

Wiz Research identified active exploitation of CVE-2026-42016, CVE-2026-42018, and CVE-2026-82329 in JFrog Artifactory between August 15 and September 8, 2026. Attackers chain the anonymous-token exposure (CVE-2026-42018) with the token scope-validation flaw (CVE-2026-42016) to obtain admin-scoped tokens, while CVE-2026-82329 allows unauthenticated administrative access in default configurations. Observed post-exploitation includes persistent administrator accounts created in under five minutes, malicious Groovy plugin deployment, ad-hoc command execution, Rust-based C2 backdoors dropped to writable paths, and webshell uploads. Wiz measured that 59-62% of organizations running Artifactory remained vulnerable to the chained CVEs weeks after disclosure, and the vulnerabilities were already included in CISA KEV.

Wiz Blog·updated Sep 14, 09:27 UTC · 2d ago·first Sep 10, 19:04 UTC · 6d agoExploit / PoC in the wild 7 sourcesCVE-2026-42016CVE-2026-42018CVE-2026-823292· 1 read

VectraRAT: An Undocumented Full-Stack MaaS Built From Scratch

SOCRadar's Threat Research Unit documents VectraRAT, a previously unreported full-stack Malware-as-a-Service platform built entirely from scratch.

SOCRadar's Threat Research Unit (STRU) documented VectraRAT, an undocumented Malware-as-a-Service platform. Unlike most MaaS offerings derived from leaked builders, VectraRAT was built entirely from scratch. The report details the platform's architecture and capabilities offered to affiliate customers.

SOCRadar·updated 14:36 UTC · 5h ago·first Sep 14, 12:50 UTC · 2d agoMalware 3 sources

JFrog security advisory (AV26-867) – Update 1

CISA added actively exploited CVE-2026-82329 in JFrog Artifactory to its KEV catalog; administrators of affected versions should patch.

Canada's Cyber Centre (advisory AV26-867, Update 1) relays a JFrog security advisory covering Artifactory, where open-source reporting indicates CVE-2026-82329 is being exploited in the wild. CISA added the flaw to its Known Exploited Vulnerabilities catalog on September 2, 2026. Multiple Artifactory release lines prior to fixed versions (e.g., prior to 7.111.21 through 7.161.20) are affected; users and administrators are urged to apply available updates.

Canadian Centre for Cyber Security·Sep 2, 18:55 UTC · 14d agoExploit / PoC in the wildCVE-2026-82329

MetroLLM-Bench: Evaluating Language Models as Transit Kiosk Runtimes

MetroLLM-Bench is a 955-case benchmark testing language models as transit kiosk tool-calling runtimes across six real metro systems.

The benchmark covers 37-414-station metro systems and eleven task categories including routing, fare calculation, disruptions, accessibility, and adversarial input, with 14 deterministic and 8 semantic scoring components. Of 26 models from six vendors, a PEFT-tuned 4B Qwen 3.5 student scored 91.3 on Tier 1, exceeding GPT-5.6 (90.6/90.0), while Muse Glimmer 30B led the composite ranking. A deterministic rule-based baseline reached 84.6, and PEFT gains over base models shrank from +7.03 points at 2B to -0.91 at 27B.

Hugging Face daily papers·Sep 8, 20:00 UTC · 8d agoAI research

The Pelican comparison grid for Astra is pretty interesting

Simon Willison's pelican SVG comparison shows GPT-6 Astra producing markedly better images than GPT-5.6 Sol, Terra, and Luna across reasoning levels.

Willison generated pelicans-riding-bicycles SVGs with newly accessed GPT-6 Astra at low through max reasoning levels and rendered them in a comparison grid against GPT-5.6 Sol, Terra, and Luna. Astra's outputs were markedly more coherent, while even the best GPT-5.6-Sol images remained largely abstract shapes. Astra does not support a reasoning=none setting, so all comparisons involved reasoning-enabled runs.

Simon Willison·Sep 4, 23:59 UTC · 11d agoAI research

VoT: Vision-of-Thought for Unified Multimodal Representation Alignment

Researchers propose Vision-of-Thought (VoT), a discrete visual-planning token layer between VLMs and diffusion transformers improving text-to-image semantic alignment.

VoT introduces a discrete visual-thinking layer between vision-language models and diffusion transformers, letting the VLM act as a multimodal planner that emits tokens describing objects and layouts before pixel generation. A specialized VoT tokenizer is trained with VLM alignment, feature reconstruction, and vector-quantization losses. Experiments show improved semantic alignment and a structured, interpretable interface for controllable generation.

arXiv cs.AI / cs.LG / cs.CL·Sep 7, 17:53 UTC · 9d agoAI research

Dependency-Check: Open-source Software Composition Analysis (SCA) tool

Help Net Security·Aug 5, 11:29 UTC · Aug 5, 2026Vulnerability1

Hunting Vulnerabilities Using Frontier Models

Okta used frontier AI models GPT-5.5 Cyber and Mythos via OpenAI and Anthropic programs to scan millions of code lines for vulnerabilities.

Okta describes using frontier AI models, including GPT-5.5 Cyber Preview (TAC) and Mythos Preview, through OpenAI's Daybreak Cyber Partner Program and Anthropic's Project Glasswing to hunt vulnerabilities across its product codebase. The team built a custom Python orchestrator with strong isolation, vendor-agnostic model support, and four distinct scanning pipelines executed as isolated Codex or Claude Code sessions with progressive context loading to reduce context bloat. Human experts and AI agents worked both autonomously and in paired hunts, and Okta reports the best results when humans and agents taught each other.

Okta Security·Sep 8, 00:00 UTC · 8d agoResearch

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

ModaLens: Measuring Image Sensitivity in Report-Conditioned Medical VLMs

ModaLens image-swap audit shows report availability cuts MedGemma-27B image sensitivity on MIMIC-CXR from 20.94% to 4.26% answer changes.

ModaLens is a paired image-swap audit measuring how report availability affects image sensitivity in report-conditioned medical VLMs. On MedGemma-27B across 3,199 paired MIMIC-CXR cases from 293 patients (14 questions per case), generated answers changed on 4.26% of image-swap trials with the report versus 20.94% without it, a 16.7-point paired difference (95% CI 15.6-17.7). The original prompt with a lowercase first-token readout gave 4.70% versus 17.07%, and the direction replicated in two further model lineages. Labels derived from reports limit conclusions about visual correctness; code, prompts, and run records are publicly released.

Hugging Face daily papers·Sep 13, 20:00 UTC · 3d agoAI research

Knowing What Not to Answer: Selective Non-Compliance in Vision-Language Models

Researchers introduce KoNA, a benchmark exposing vision-language models' failures at selective non-compliance, plus fine-tuning that improves refusal and abstention accuracy.

KoNA is a benchmark for evaluating selective non-compliance in vision-language models across five categories: False Premise, Visual Inaccessibility, Universal Unknown, Task Feasibility and Safety. It tests both query-level and component-level non-compliance using paired single and compound queries, and evaluations across diverse VLMs show models often fail to refuse, correct or abstain appropriately, with failures worsening on compound queries. Fine-tuning VLMs on KoNA examples substantially improves non-compliance accuracy while largely maintaining performance on fully answerable tasks.

Hugging Face daily papers·Sep 3, 20:00 UTC · 13d agoAI research1