Search: “lpe”

Cognition's SWE-2 achieves 92.8 on Terminal-Bench 2.1

Cognition releases SWE-2, a 2.8T-parameter MoE coding model post-trained from Kimi K3, scoring 92.8 on Terminal-Bench 2.1.

SWE-2 is a proprietary mixture-of-experts model with 2.8T total parameters and 104B active per token, built on the Kimi K3 base with additional Cognition reinforcement-learning post-training for agentic coding. Vendor-reported benchmarks include FrontierCode 1.1 Main 50.0, DeepSWE 1.1 73.0, Terminal-Bench 2.1 92.8, and Terminal-Bench 4.0 27.3. It claims to be one point behind Claude Fable 5.1 on FrontierCode at a claimed 64% lower cost, but trails Fable 5.1 and GPT-6 Astra by a wide margin on long-horizon Terminal-Bench 4.0 tasks. The model is available today in Devin Desktop and CLI, with no published weights, no per-token API pricing, and all figures pending independent replication.

Hacker News · AI·updated 13:01 UTC · 7h ago·first Sep 10, 16:52 UTC · 6d agoModel release 11 sourcesHN 40↑ · 18 comments

Linux Detection Engineering - Local Privilege Escalation

Elastic details a layered detection framework for Linux local privilege escalation, covering 2026's copy-on-write bug wave and LLM-assisted discovery.

Elastic Security Labs describes how most Linux local privilege escalations share a common host flow — an unprivileged process launched from a writable path becoming root — and proposes layered detections combining general outcome-based rules with per-technique rules in Elastic Defend and Auditd. It tracks 13 recent LPE disclosures, seven of which share a copy-on-write/zero-copy bug class, including Copy Fail, DirtyFrag, Fragnesia, DirtyDecrypt, DirtyClone, pedit COW, and RefluXFS. Qualys attributes RefluXFS to an LLM-assisted research effort with Anthropic using Claude Mythos Preview, and another bug is credited to an LLM-assisted workflow. Detection and endpoint rules are published in Elastic's detection-rules and protections-artifacts repositories.

Elastic Security Labs·Sep 11, 00:00 UTC · 5d agoResearch

CVE-2026-86287: Net::IP::LPM versions before 1.12 for Perl accept malformed prefix lengths

CVE-2026-86287: Net::IP::LPM Perl library before 1.12 accepts malformed IP prefix lengths, risking incorrect longest-prefix-match results.

Net::IP::LPM versions before 1.12 for Perl accept malformed prefix lengths, per an oss-security disclosure posted September 7, 2026. The lax input validation in this longest-prefix-match library could produce incorrect matching behavior. The fix is available in version 1.12 on CPAN.

oss-security·Sep 7, 18:40 UTC · 9d agoVulnerabilityCVE-2026-86287

Local Privilege Escalation (LPE) in FolkPatch due to Hardcoded Default SuperKey

A hardcoded default SuperKey in FolkPatch, an APatch-based kernel patching tool, enables local privilege escalation limited to FolkPatch's downstream code.

A disclosure posted to the oss-security mailing list describes a local privilege escalation (LPE) in FolkPatch caused by a hardcoded default SuperKey. FolkPatch is a downstream project based on APatch that utilizes its own custom KernelPatch. According to the post, the vulnerability is specific to FolkPatch's downstream modifications rather than upstream APatch or KernelPatch code.

oss-security·Sep 12, 14:32 UTC · 4d agoVulnerability1

Real-SWE: Benchmarking AI models on private, real-world, enterprise codebases

Real-SWE benchmark tests coding agents on licensed private enterprise codebases; top model Fable 5.1 resolves only 38.8% of tasks.

Real-SWE is a new benchmark evaluating frontier AI coding agents on tasks drawn from private production codebases licensed from real companies, spanning billing, tax calculation, and cross-service migrations. Fable 5.1 with Claude Code leads at 38.8% resolution rate (pass@1 over eight runs), followed by GPT-6 Astra Codex CLI at 33.8% and Gemini 3.8 Flash Gemini CLI at 31.2%. Tasks use native harnesses and realistic tooling including Docker, Kubernetes, PostgreSQL, Redis, and Linear; median reference solutions edit 11 files versus 6 for DeepSWE and FrontierCode.

Hacker News · AI·Sep 12, 20:25 UTC · 3d agoAI researchHN 45↑ · 31 comments

Metasploit Wrap Up: Lot of summer shells and fit http profiles

Metasploit 6.5 ships 13 new modules including RCEs for SonicWall SMA1000, Ghost CMS, Joomla JCE and Langflow, plus a Linux kernel LPE.

Rapid7's Metasploit wrap-up covers the 6.5 release, which adds 13 new modules, including remote code execution exploits for WordPress WP2Shell, Ghost CMS, Joomla JCE, Langflow, OpenCATS, Pterodactyl Panel, SonicWall SMA1000, Ray Dashboard and a Pix-for-WooCommerce plugin. The Fragnesia Linux kernel local privilege escalation (CVE-2026-46300) also landed. The release introduces HTTP malleable profiles, MCP functionality, Linux multi-fetch payloads and improved Windows on ARM support.

Rapid7 Blog·Aug 14, 21:27 UTC · Aug 14, 2026ToolsCVE-2026-46300

Plug 'n' Pray: Agentic LLM-based Detection of Potential Log File Exposures in Third-Party Content Management System Plugins

Agentic LLM analysis validates 79 log file exposures across 62 of the 300 most-installed WordPress plugins, covering 250M+ active installations.

Researchers built an agentic LLM-based framework combining static and dynamic analysis to automatically detect insecure log files created by WordPress plugins. Scanning the 300 most-installed plugins, which account for roughly 75% of all active installations in the official ecosystem, it produced 81 findings with 79 manually reproduced across 62 plugins. Insufficiently secured log files can disclose credentials and personal data and have led to website compromises. The authors derive a taxonomy of log path and protection patterns and best practices, finding multi-layered protection often absent.

arXiv cs.CR·Sep 15, 13:29 UTC · 1d agoResearch

China-Linked Hackers Exploit Chrome and Windows Zero-Days in BlueMoon Attacks

China-linked clusters deploy the BlueMoon kit chaining Chrome V8 CVE-2026-85046 and Windows LPE CVE-2026-85880 in espionage campaigns.

Proofpoint researchers identified BlueMoon, an exploit kit combining a V8 type-confusion RCE (CVE-2026-85046), a V8 sandbox escape, and a Windows kernel privilege-escalation flaw (CVE-2026-85880), first observed August 28, 2026. At least four clusters adopted it, led by TA412 (also tracked as APT31/Violet Typhoon) and followed by UNK_LateNight, UNK_DoubleCheck, and UNK_QuietRacket, targeting aerospace, manufacturing, government, consulting, and financial sectors. The kit exploited a nearly four-week patch gap between the public Chromium commit (August 7) and stable rollout (September 3). TA412 delivered the GemStone Chrome extension masquerading as a Gemini companion, while other clusters deployed ShadowPad via DLL sideloading, a Rust loader, and DoH-based C2.

GBHackers·updated 17:13 UTC · 2h ago·first Sep 10, 05:15 UTC · 6d agoExploit / PoC in the wild 19 sourcesCVE-2026-85046CVE-2026-858801

Who Should Grade My Work? Student Perspectives on Transparent AI-Assisted Writing Assessment in Higher Education

A Saudi university study finds students value ChatGPT writing feedback but treat human instructors as the final grading authority.

Thirteen male undergraduate computing students at a Saudi public university completed handwritten writing tasks that were scored by ChatGPT using a rubric-based prompt, then reflected after being told the score and feedback were AI-generated. Inductive thematic analysis identified four themes: perceived feedback usefulness, awareness of AI's contextual and pedagogical limitations, conditional trust, and reflection on the instructor's institutional role. Participants accepted GenAI feedback for surface-level revision but consistently positioned human instructors as the authority over grading decisions, distinguishing feedback utility from evaluative authority.

arXiv cs.AI / cs.LG / cs.CL·Sep 4, 16:49 UTC · 12d agoAI research

Four Spy Groups Used the Same Chrome and Windows Exploit Kit Within a Week

Proofpoint links four espionage clusters, including China's APT31, using shared exploit kit BlueMoon chaining Chrome V8 and Windows ALPC zero-days.

Proofpoint reports a previously undocumented exploit kit, BlueMoon, chains Chrome V8 type confusion CVE-2026-85046 with an unassigned V8 sandbox escape and Windows ALPC heap overflow CVE-2026-85880 to achieve code execution and local privilege escalation. APT31 first used it on August 28, 2026 against US NGOs, mining, and commodity trading firms, deploying the GemStone browser backdoor disguised as a Google Gemini extension via the GhostChrome-X integrity bypass. UNK_LateNight, UNK_DoubleCheck, and UNK_QuietRacket followed on September 2-3, targeting US aerospace, Vietnamese manufacturing, and Indonesian/Singaporean sectors with ShadowPad and sideloaded Rust and .NET payloads. Both V8 flaws were patch-gap zero-days, and verbose code suggests possible AI-assisted development.

The Hacker News·Sep 9, 16:34 UTC · 7d agoThreat actor in the wild 2 sourcesCVE-2026-85046CVE-2026-858801

Cognition Releases SWE-2: A Kimi K3 Post-Trained Coding Model That Matches Fable 5.1 on FrontierCode at 64% Lower Cost

Cognition released SWE-2, an RL post-trained coding model from Kimi K3, scoring 50.0% on FrontierCode 1.1 Main and available only inside Devin.

Cognition released SWE-2, its most capable coding model, post-trained with reinforcement learning from Moonshot AI's 2.8T-parameter Kimi K3 base. It scores 50.0% on FrontierCode 1.1 Main, within 1 point of Fable 5.1 at 64% lower cost, and RL reportedly adds 5-6 points over the K3 base on many benchmarks. It is the first Cognition model with selectable reasoning-effort levels all trained in a single RL run using Pareto-slope-matched cost penalties. There are no open weights and no standalone API; it runs only inside Devin (Desktop, CLI, with Web and Fusion rolling out), free for paid tiers through October 10, 2026.

MarkTechPost·Sep 12, 23:56 UTC · 3d agoModel release1

Lazarus hackers pair fake job offers with Windows zero-day exploit

Lazarus' Operation Dream Job targets the defense sector with fake job offers, a Windows LPE zero-day (CVE-2026-68820), and new Troy and RelayShell backdoors.

Check Point documented two parallel infection chains in Lazarus' Operation Dream Job: a digitally signed PDF viewer with DLL sideloading delivering the MISTPEN in-memory downloader, and a trojanized SecurityPDF viewer, distributed via fake Enveil websites, installing the newly documented Troy backdoor. A component exploits CVE-2026-68820, a local privilege escalation zero-day in the Windows AFD.sys driver used since at least early July, which Microsoft patched on August 11, 2026, enabling deployment of the FudModule kernel-mode rootkit. Lazarus also compromised Roundcube webmail servers vulnerable to CVE-2025-49113 to deploy the RelayShell PHP web shell and relay C2 traffic, targeting defense-sector organizations primarily in Western Europe and India. Separately, CERT-UA documented Sandworm's UAC-0145 using fake job offers against IT professionals since May 2026.

Help Net Security·Aug 12, 00:00 UTC · Aug 12, 2026Threat actor in the wildCVE-2026-68820CVE-2025-49113

Emergency Department Revisit Quality Review Screening: Exploring Human Decision-Making and Artificial Intelligence Support

A retrospective study found GPT-4 over-flagged emergency department revisit cases while an LLM knowledge-graph screener achieved 83-100% positive predictive value.

In an exploratory retrospective study of 99 emergency department diagnosis pairs from a multihospital health system, clinicians and GPT-4 independently judged whether revisit pairs warranted further assessment. GPT-4 responses correlated poorly with clinicians, flagging 94% of pairs for follow-up, 4.4-13.3 times more than clinicians, though prompt engineering was minimal. An algorithm leveraging an LLM-populated knowledge graph (KGA) achieved 83-100% positive predictive value against at least one clinician rater, suggesting LLM-based screening could broaden revisit quality review without substantially increasing reviewer workload.

arXiv cs.AI / cs.LG / cs.CL·Sep 9, 16:37 UTC · 7d agoAI research1

Apple macOS Screen Sharing Flaw Exploited on Internet

Actively exploited CVE-2026-65400 in macOS Screen Sharing grants pre-auth root access; attackers deploy Monero miners on exposed systems.

NCSC-NL reported active exploitation of CVE-2026-65400 (CVSS 9.8), an authentication flaw in macOS Screen Sharing patched on August 6, 2026, with root access gained and a Monero miner planted on internet-exposed systems using port 5900. Related Screen Sharing bugs CVE-2026-43779, CVE-2026-43777, and CVE-2026-43760 were fixed in macOS Tahoe 26.6, and researcher @osxreverser noted a pre-auth flaw fixed alongside them affecting roughly 40,000 exposed hosts. Calif said an AI agent produced working exploits for both pre-auth bugs in four hours, underscoring the shrinking gap between patch release and weaponization.

The Hacker News·Aug 19, 05:22 UTC · 28d agoExploit / PoC in the wildCVE-2026-65400CVE-2026-43779CVE-2026-43777+1 CVEs

Necessary or Sufficient? Evaluating LLM Explanations With Behavioural Evidence

New framework tests whether LLM-cited explanation factors are necessary or sufficient, finding weak correlation across Claude, GPT, and Gemini models.

An arXiv paper introduces black-box intervention tests measuring whether factors LLMs cite in their explanations are necessary or sufficient for their outputs in agent oversight workflows. Across eight models from the Claude, GPT, and Gemini families, Spearman correlations between cited rankings and measured influence ranged from 0.349-0.354 (advisor recommendation) to 0.431-0.580 (prompt monitoring). Uncited factors scored above the lowest cited factor in up to 57.6% of advisor responses, showing cited top-three factors do not reliably identify the most influential inputs.

arXiv cs.AI / cs.LG / cs.CL·Sep 4, 17:37 UTC · 12d agoAI safety & security1

UAT-10147 Uses AI to Scale Server Attacks, Deploys SPECTRE With EDR Bypass and Linux Rootkit

Cisco Talos exposes UAT-10147, a Chinese-speaking group using AI tools to automate intrusions, deploy SPECTRE, BadIIS, and rootkits against web servers worldwide.

Cisco Talos detailed UAT-10147, a Chinese-speaking cybercrime group conducting SEO fraud and data theft against Windows and Linux web servers in education, media, technology, and gaming sectors, with most victims in Brazil, Bolivia, China, Canada, and Vietnam. The actor exploits publicly disclosed vulnerabilities for initial access, including Zimbra (CVE-2022-27925) and Alibaba Nacos (CVE-2021-29441), and abuses Linux LPE flaws like CVE-2022-0847 and CVE-2021-3156 for root. Its toolset includes AI-assisted frameworks DeepAudit and PentestGPT, plus implants such as SPECTRE, BadIIS, Quasar RAT, Gh0stCringe, and Noodle RAT. An exposed directory contained a target list of roughly 170,000 URLs, with the US, India, UK, Germany, and Netherlands as top destinations.

The Hacker News·Aug 27, 18:33 UTC · 20d agoThreat actor in the wildCVE-2022-0995CVE-2021-3156CVE-2015-5287+8 CVEs

I accidentally turned LLM memory into program analysis

A pwning.systems write-up describes how LLM memory functionality was unexpectedly repurposed into a program analysis technique.

A security research post on pwning.systems describes the author's discovery that LLM memory behavior effectively functioned as program analysis. The write-up is hosted on a security-focused blog and surfaced via a security-tagged link aggregator. Detailed technical content is not included in this feed, limiting verifiable specifics.

Lobsters · security·Aug 29, 03:16 UTC · 18d agoResearch1

⚡ Weekly Recap: Rogue AI Models, $88M Bitcoin Theft, Water

The Hacker News·Aug 3, 14:03 UTC · Aug 3, 2026Data breachCVE-2026-42897CVE-2026-66066

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

Lazarus Exploits Windows Zero-Day to Gain SYSTEM Access and Deploy Backdoor

Lazarus exploits Windows AFD.sys zero-day CVE-2026-68820 in Operation Dream Job to deploy Troy backdoor at defense firms.

Check Point attributes Operation Dream Job attacks to Lazarus Group exploiting CVE-2026-68820 (CVSS 7.0), a privilege escalation flaw in Windows AFD.sys patched in August 2026 Patch Tuesday. The campaign targets defense and aerospace firms in France, Germany, Brazil, and India via trojanized PDF viewers and DLL side-loading, deploying backdoors Troy, ForestTiger, and the FudModule 3.1 kernel rootkit. Attackers also compromise WordPress, SharePoint, and Roundcube servers as C2, using CVE-2025-49113 and the RelayShell PHP web shell.

The Hacker News·Aug 15, 00:00 UTC · Aug 15, 2026Threat actor in the wildCVE-2026-68820CVE-2025-491132

Bag of Tricks or Bag of Myths? Reducing Modeling Complexity with Task Knowledge in Explainable Suicide Risk Assessment

Researchers audit 31 NLP techniques for clinician-annotated suicide risk prediction, finding only 5 of 31 comparisons yield reliable gains.

A study of 1,635 clinician-annotated social media posts ran roughly 300 controlled experiments across 7 methodological families, auditing techniques such as model scaling, synthetic data, ensembling, and threshold tuning under severe class imbalance. The proposed system reformulates risk factor prediction as entailment between posts and codebook definitions, using architecturally diverse ensembles with class-balanced training and deployment-consistent calibration. It scores 0.8203 for risk, 0.7953 for evidence, and 0.7045 macro-F1 for factors, ranking third among 53 teams.

arXiv cs.AI / cs.LG / cs.CL·Sep 7, 17:14 UTC · 9d agoAI research

Parallels Desktop Vulnerability Lets Non-Admin Mac Users Execute Code as Root

JFrog researchers disclosed CVE-2026-90894, a critical Parallels Desktop local privilege escalation letting non-admin Mac users execute code as root; fixed in 27.0.0.

JFrog researchers found that Parallels Desktop's privileged prl_disp_service daemon on macOS accepts unsigned local clients through a world-writable socket and allows argument injection into a tar command during appliance installation. Injecting the --use-compress-program option makes tar execute an attacker-controlled program as root, and a one-line proof of concept yielded a root shell without needing a running virtual machine. The flaw, dubbed ParaShells, was confirmed in Parallels Desktop 26.4.0 build 57513 on Apple Silicon and is fixed in version 27.0.0. No exploitation in the wild has been reported; administrators should inventory and upgrade affected installations and restrict local account access.

Cyber Security News·updated 13:14 UTC · 6h ago·first 09:24 UTC · 10h agoVulnerability 4 sourcesCVE-2026-90894

Coding Agents Have Converged: Why the SWE-bench Leaderboard Can No Longer Order Its Top Entries, and What to Measure Instead

Audit of 254 SWE-bench submissions finds top coding-agent entries statistically inseparable, so small leaderboard gaps no longer establish rank.

The paper audits 254 SWE-bench submissions across four splits without running models. On Verified, the top two entries each resolve 396 of 500 instances, and exact paired McNemar tests separate none of the 29 adjacent top-thirty pairs at alpha=0.05. Within-model scaffold score ranges reach 29.8 percentage points, versus an 8.8-point spread among the top thirty. The authors release a five-step audit protocol and recommend reporting comparison-set-specific resolution and model-scaffold provenance.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 16:29 UTC · 1d agoAI research

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

Not All Prompts Are Equal: Exploration-Guided Prompt Scaffolding for Multimodal Reinforcement Post-Training

Exploration-guided prompt scaffolding rewrites training prompts by Exploration Potential Score, boosting multimodal RL post-training accuracy up to 11.5%.

The paper proposes dynamically adapting the training prompt distribution during online RL post-training of multimodal LLMs using the Exploration Potential Score (EPS), a lightweight rollout-based proxy for prompt utility computed from on-policy statistics with no additional overhead. Rather than discarding low-utility prompts, a teacher model generates scaffolded rewrites that preserve task intent while making training more informative. Integrated with GRPO on Geo3K and MMK12, the method achieves up to 9.7% relative in-domain improvement plus 11.5% on MathVision and 11.1% on MMMU-Pro.

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

SWE-Bench Pro Verified: A Reliable Benchmark for Software Engineering Agents

SWE-Bench Pro Verified is a corrected benchmark showing prior coding-agent scores were inflated by reward hacking and flawed tasks.

Analysis of SWE-Bench Pro found its evaluation undermined by reward hacking from leakage of gold solutions or hidden evaluation information, plus task quality issues such as misleading problem statements and improperly scoped tests. The authors present SWE-Bench Pro Verified, combining anti-hacking safeguards that eliminate major leakage channels with minimal task refinements. Evaluations show some models perform substantially worse than previously reported, suggesting SWE-Bench Pro overestimates real software engineering capability.

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

From Reweighting to Rewriting: Unlocking the Intervention Effects of Influential Samples in Training Data Attribution

Study shows rewriting responses of influence-selected training examples shifts LLM behavior more strongly than reweighting the same samples.

The paper examines training data attribution, arguing that influence functions identify high-leverage examples whose value goes unrealized under conventional weight-based reweighting interventions. It introduces influence-guided response rewriting, which replaces the responses of influence-selected examples with behavior-aligned or behavior-opposed supervision while keeping instructions fixed, tested across four open-weight LLMs using epistemic abstention as the primary testbed. Rewriting produces stronger, more persistent, and bidirectional behavioral shifts, including on safety refusal, while reweighting the same examples yields weak, inconsistent effects. The results motivate intervention-aware evaluation of TDA methods.

Hugging Face daily papers·Sep 1, 20:00 UTC · 15d agoAI research

Engineered Persuasion: Evaluating Personalized Pretexts in LLM-Generated Spear Phishing

A study of 180 US workers found each LLM phishing personalization level raised click-intention odds by 28%, but credibility depends on context fit.

The arXiv paper evaluates how personalized pretexts in LLM-generated spear phishing affect perceived credibility, using 180 US working adults across 1,436 evaluations of emails with four cumulative personalization levels, from workplace context to shared-project details. Convincingness rose 2.40 points per level in sensitivity analysis and click-intention odds increased 28% per level, while non-clickers shifted toward deleting rather than reporting. Qualitative coding showed details matching the recipient's role and routines supported credibility, whereas incorrect, vague, or channel-inappropriate details raised suspicion. The authors argue personalization effectiveness depends on pretext fit, with implications for workplace security training.

arXiv cs.CR·Sep 3, 19:21 UTC · 13d agoResearch

One Symptom, Three Levers: A Critical Review of On-Policy Self-Distillation

A review paper frames on-policy self-distillation collapse as governed by three levers: token weighting, privileged information, and guidance decay.

The paper critically reviews On-Policy Self-Distillation (OPSD), where a language model trains on its own generations scored token-by-token by a teacher conditioned on privileged information such as reference solutions or environment feedback. It identifies collapse, the progressive narrowing of producible reasoning paths, as the dominant failure mode and analyzes it through three levers: signal weighting, the nature of privileged information, and teacher dynamics. The review is restricted to mathematical reasoning, reports no new experiments, and offers a shared vocabulary separating settled findings from disputed ones.

Hugging Face daily papers·Aug 25, 20:00 UTC · 22d agoAI research

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

Root-Cause Attribution Is a Search Problem: Continual Search for Long-Horizon Agent Failures

Continual Search framework iteratively prompts LLM judges to keep searching agent execution logs, boosting long-horizon failure root-cause attribution accuracy.

The paper frames automated root-cause attribution (RCA) for long-horizon AI agent failures as a search problem, since relevant evidence is sparse and distributed across massive execution traces. The authors propose Continual Search, an iterative framework that nudges an LLM judge across successive turns to keep hunting unresolved diagnostic evidence instead of settling on an early plausible diagnosis. They introduce MegaRCA-Mix, a benchmark of 50 human-annotated failure trials on long-horizon, execution-heavy tasks. On MegaRCA-Mix, Continual Search improves GPT-5.5's F1 from 0.349 to 0.498 (over 40% gain), and lower-tier models can surpass higher-tier counterparts when search is effective.

Hugging Face daily papers·Sep 10, 20:00 UTC · 6d agoAI research1

ActGuard: Pre-execution Action Auditing against Indirect Prompt Injection in LLM Agents

ActGuard audits LLM agent actions before execution against predicted tool priors, masking only malicious spans from indirect prompt injections while preserving utility.

ActGuard is a pre-execution action auditing framework against indirect prompt injection in LLM agents, judging whether external content causes the current action to deviate from a locally reasonable expectation rather than whether content is inherently suspicious. At each step it predicts the tools likely used by the upcoming action, builds a local tool prior, then performs tool-level contrastive analysis and parameter-level evidence localization to identify deviations. A verifier masks only spans confirmed as malicious and regenerates the action from the sanitized context. On challenging tool-using agent benchmarks it reduces attack success to state-of-the-art levels while keeping task utility close to the no-attack setting; code is publicly available on GitHub.

arXiv cs.CR·Sep 14, 03:47 UTC · 2d agoAI safety & security

ActReview: Rebuttal-Guided Training Data and Rubric Rewards for Actionable Peer Review Generation

ActReview post-trains Qwen3-8B-Base on OpenReview rebuttals to generate actionable peer-review feedback with grounded revision suggestions, benchmarked on 1,000 curated instances.

The paper defines Actionable Peer-review Generation as diagnostic claim generation plus revision suggestion generation and introduces ActReview, a rebuttal-guided post-training framework. From OpenReview review-rebuttal threads the authors build ActReview-40K, aligning reviewer weaknesses with author responses grounded in localized paper evidence, and post-train Qwen3-8B-Base with multi-task SFT followed by GRPO using weakness-specific rubric rewards. They also release ActReview-Bench, a human-curated 1,000-instance benchmark, on which ActReview outperforms prior specialized review-generation models on actionability and grounding while remaining competitive with strong prompt-based LLMs. Human evaluation confirms improved revision usefulness but identifies a remaining gap in technical accuracy.

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

Toward an Empirical Probabilistic Risk Manifestation Model of Organizational Cybersecurity in SMEs

Empirical study of 22 SME security assessments builds a probabilistic risk model and shows assessments can be cut 24-45% while retaining most critical findings.

Researchers analyzed 281 validated security findings from 22 real-world SME cybersecurity assessments conducted over two years via a pro bono university clinic. They derived an empirical Risk Manifestation Model linking eight organizational security functions to two exposure conditions, five attack mechanisms, and six outcome categories, using probability propagation to identify dominant risk pathways. The dominant pathway runs from asset exposure to credential compromise to unauthorized access, stable under leave-one-organization-out analysis. Retaining six functions reduces assessment burden by 24% while preserving 97% of critical findings; five functions cut burden 45% while preserving 89% of critical findings.

arXiv cs.CR·Sep 14, 01:15 UTC · 2d agoResearch

ExecCritic: Learn to Test, Test to Improve for Coding Agents

ExecCritic separates test generation from patching for coding agents, lifting SWE-bench Verified resolution to 72.6%.

ExecCritic pairs a test-verify-revise scaffold with role-specific reinforcement learning: a Test agent writes repository-native tests and a Repair agent fixes code from execution feedback, both using Qwen-3.5-35B-A3B backbones. Post-trained Qwen agents compose to 72.6% on SWE-bench Verified, an 11.4-point gain over the 61.2% no-test baseline, without stronger-model or oracle feedback at evaluation time. The work shows test quality is the key variable: base-agent tests lowered resolution to 57.3% while GPT-5.6-sol tests raised it to 65.3%.

arXiv cs.AI / cs.LG / cs.CL·Sep 8, 17:53 UTC · 8d agoAI research1

Re: Retrospective by 'gpg.fail' authors

oss-security follow-up to a gpg.fail authors' retrospective argues that GnuPG's unmaintained, forgotten code is a security liability and should be deleted.

On the oss-security mailing list, Soatok Dreamseeker responds to a retrospective published by the gpg.fail authors, asking why unmaintained code is retained and stating that forgotten, unmaintained code is a liability and effectively unmaintained by definition. The visible reply addresses Werner Koch and concerns GnuPG code maintenance. No specific CVE, flaw, or exploitation is described in this snippet.

oss-security·17:16 UTC · 2h agoVulnerability 8 sources1

Reference-Based Bias Detection in LLMs via Relative Representations of Hidden States

Researchers propose auditing LLM bias via relative hidden-state representations, detecting bias increases with 3-50x less compute than output-level benchmarks.

The paper introduces a reference-based bias auditing method that compares hidden-state representations across model variants, such as before and after fine-tuning, by encoding sentences relative to a fixed anchor set. The resulting Representational Bias Shift (Delta-B) correlates with output-level bias change in 15 of 18 tested settings, reaching |r| = 0.84 under full fine-tuning across WildGuardMix, DecodingTrust, and ToxiGen benchmarks. Thresholding Delta-B detects checkpoints whose bias increased with ROC AUC between 0.65 and 0.99 and beats a SEAT-based baseline, while auditing a model in about three minutes with 3-50x less compute.

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

What Does an LLM-Agent Leaderboard Rank Actually Compare?

A methodological study shows close LLM-agent leaderboard rank gaps on SWE-bench and similar benchmarks often do not support superiority claims.

The paper defines an estimand-aware pairwise procedure for comparing agents, checking common support and applying explicit uncertainty rules and practical margins. Across SWE-bench, AgentRewardBench, and tau2-bench, close rank differences are frequently unresolved, and proxy labels or utility rules can change which system is selected. The authors argue a leaderboard score summarizes a released evaluation but does not by itself justify pairwise superiority conclusions.

arXiv cs.AI / cs.LG / cs.CL·Sep 7, 17:31 UTC · 9d agoAI research

CVE-2026-50522: Microsoft SharePoint Server RCE Vulnerability

CVE-2026-50522 is an unauthenticated .NET deserialization RCE in SharePoint Server Subscription, 2019, and 2016; Microsoft patched it July 14, 2026.

CVE-2026-50522 (CWE-502) lets unauthenticated attackers send crafted serialized .NET payloads to network-accessible SharePoint endpoints, triggering gadget-chain deserialization and remote code execution under the privileged SharePoint service account. It affects SharePoint Server Subscription Edition, SharePoint Server 2019, and SharePoint Server 2016 Enterprise; Microsoft published the advisory on July 14, 2026. The EPSS score is 20.346%, but no public PoC exists and the flaw is not in CISA's KEV catalog at publication. Mitigations include applying Microsoft updates, rotating SharePoint machine keys, enabling AMSI, and restricting internet exposure.

Web discovery (articles for new exploits & KEV entries)·Sep 7, 19:42 UTC · 9d agoVulnerability 3 sourcesCVE-2026-505221

τ^τ-Bench: An Environment for End-To-End, Realistic Agent Construction

New τ^τ-bench tasks coding agents with building deployable customer-service agents; best config, Claude Opus 5, passes only 23.9% of simulations.

Researchers introduce τ^τ-bench, an end-to-end benchmark where a developer agent must build a complete customer-service agent from real business records, a client with requirements, a production API, an inherited codebase, and cost/model limits, then is scored by deploying it against held-out simulated users. Across 53 tasks in four domains, the strongest configuration, Claude Opus 5 under Claude Code, passes just 23.9% of evaluation simulations versus an 82.2% expert-authored reference ceiling. Failure modes mirror those of human developers: shallow queries instead of deep record comprehension, almost no client communication, and shipping the first architecture that runs rather than experimenting.

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