ZeroHour
DashboardNewsCVEsExploitsIncidentsAIBriefAnalyticsEntitiesWatchlistIOCsAsk the feedSourcesFeeds & APIKiosk mode

Search: “memory-safety”

40 stories

Searchkeywordsmart
DomainAllCyberAIWindow24h3d7d30dAllIn the wildImportant onlyShow duplicates
CategoryAllVulnerability 19Exploit / PoC 5Research 3Advisory 1Phishing & fraud 1Policy & legal 1
SeverityAllcritical 7high 7medium 12low 0info 4

Smart search ranks by meaning as well as keywords (one row per story, last 45 days).

CHERI-D Reincarnate: efficient multicore CHERI temporal memory safety through allocation reincarnation (draft version)

CHERI-D Reincarnate adds allocation reincarnation to CHERI, quarantining generation IDs instead of memory to cut temporal memory-safety overheads on multicore systems.

The proposal extends CHERI-D temporal memory safety by quarantining exhausted generation IDs rather than memory slots, allowing freed memory immediate reuse through allocation reincarnation. It adds coherent ID caching and multicore support connecting physical coherence events to the virtually addressed ObjID buffer. The design was implemented as a hardware-software co-design across CHERI-Toooba FPGA softcore, QEMU, LLVM/Clang, and CheriBSD. Across evaluated workloads it substantially reduces memory-sweep frequency and quarantine overhead with low performance and hardware cost.

arXiv cs.CR · 6d agoResearch

Memory-safety defects in the upstream (abandoned) AOSP OpenCORE AAC decoder, shipped unpatched by Samsung TizenRT

Abandoned AOSP OpenCORE AAC decoder has out-of-bounds-write and wild-pointer flaws reachable from attacker-controlled frames, unpatched in Samsung TizenRT.

The OpenCORE AAC decoder in AOSP's abandoned external/opencore tree contains memory-safety defects of the out-of-bounds-write and wild-pointer class. The code is still vendored and built by multiple projects, most notably Samsung's widely deployed TizenRT embedded RTOS. The defects are reachable from untrusted media because an AAC frame is attacker-controlled. The researcher is requesting a CVE ID for the issue.

oss-security · 6d agoVulnerability1

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 · 6d agoVulnerability

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 · 8d agoAI safety & security

Fortunate Recall: Ontology-Driven Memory Lifecycle Management for Persistent Coherence in LLMs

Fortunate Recall introduces ontology-based lifecycle policies for LLM memory, cutting confabulation roughly in half (e.g., 45.1% to 22.4%) versus Mem0.

Fortunate Recall (FR) is a composable policy layer that classifies personal facts into a 10+1 behavioral ontology and applies category-specific lifecycle rules including differential temporal decay, slot-key supersession, event-time validity, and retrieval routing. FR-Bank scores 76.9% on the new 516-question LifecycleBench, ahead of Mem0, A-MEM, Memory-R1, and MemoryOS (61%-70.5%), and 75.2% on LongMemEval-S. End-to-end, confabulation drops from Mem0's 45.1% to 22.4% over answered queries, with the ranking replicating on open-weight Kimi K2.5 and transferring to the independent BEAM benchmark (46.8% vs 32.9%).

arXiv cs.AI / cs.LG / cs.CL · 6d agoAI research

Wireshark 4.6.8 patches 28 security bugs, nine in file parsers

Wireshark 4.6.8 fixes 28 security bugs, including nine crash-prone capture file parsers, misdecoded 5G fields, and memory-safety issues.

Wireshark 4.6.8 fixes 28 security bugs spanning advisories wnpa-sec-2026-64 through wnpa-sec-2026-91, including nine crash bugs in file parsers such as pcapng, Endace ERF and Tektronix K12xx that trigger when opening capture files. Fixes cover dissectors for RDP, SSH, Kerberos, H.245, CMS, C12.22 and several Bluetooth protocols, plus unnumbered memory-safety issues like a stack buffer overflow in the K12/RF5 writer. The release also corrects eight misdecoded 5G NAS/5GSM fields and moves the Unix extcap path to /usr/libexec/wireshark/extcap.

Help Net Security · Aug 13, 2026Advisory

Google Chrome 153 Released With 230 Security Fixes and Zero-Day Exploited in the Wild

Google released Chrome 153 with 230 security fixes, patching CVE-2026-87491, a V8 out-of-bounds write zero-day actively exploited in the wild.

Google shipped Chrome 153 (153.0.8010.36/.37) for Windows, macOS, and Linux with 230 security fixes, including CVE-2026-87491, a medium-severity V8 out-of-bounds write being exploited in the wild. The release also fixes five critical vulnerabilities: four in WebGL (two use-after-frees, an out-of-bounds write, a buffer overflow) and one use-after-free in Cast. High-severity V8 use-after-free and type-confusion defects are among the fixes, and Google has restricted technical details until most users are patched. Administrators should verify deployed versions on managed endpoints, as Chrome-derived browsers may follow different patch schedules.

GBHackers · 7d agoExploit / PoC in the wildCVE-2026-87491CVE-2026-87464CVE-2026-87488+19 CVEs1

MeClear: Cooperative Game-Theoretic Attribution and Risk-Aware Memory Clearance for Long-Horizon LLM Agents

MeClear uses cooperative Shapley attribution to clear harmful memories from long-horizon LLM agents, boosting task recovery by 25.5 points over baselines.

The paper introduces MeClear, a task-conditioned memory clearance framework for long-horizon LLM agents that identifies and selectively suppresses memories with negative downstream utility without permanently altering the persistent memory bank. It combines Leave-One-Out screening with sampled cooperative Shapley attribution to distribute utility across interacting evidence, resolving redundant conflict masking that single-removal evaluations miss. Across ten long dialogue memory pools it achieves 85.9% target recall and 82.3% overall task recovery, a 25.5 percentage-point improvement over LOO baselines.

arXiv cs.AI / cs.LG / cs.CL · 7d agoAI research

CISA review makes the case for eliminating vulnerability classes

CISA's FY2024-2025 vulnerability review urges eliminating recurring vulnerability classes, finding 41.5% of 2025 KEV entries map to persistent 'stubborn weaknesses'.

CISA's Vulnerability Review for fiscal years 2024 and 2025 found most compromises stemmed from opportunistic exploitation of known, exposed vulnerabilities rather than nation-state zero-days or advanced tradecraft. The review found 41.5% of 2025 KEV catalog entries map to 'stubborn weaknesses' - injection flaws, improper input validation, memory-safety failures, path traversal and broken access control - recurring on the CWE Top 25 since 2019. CISA advocates Secure by Design practices, vulnerability class elimination, and buyer-driven 'Secure by Demand' contract requirements, while warning that AI-enabled vulnerability discovery will soon accelerate exploitation.

Help Net Security · 15d agoPolicy & legal

Safe Meta-Reinforcement Learning via Information Space Reachability

Safe meta-RL framework reasons about safety in information space, learning a safety value function used for safety filtering and constrained policy optimization.

The paper proposes safe meta-RL that reasons about safety in information space, capturing both physical state and the agent's belief over the underlying task. A safety value function measures the probability of avoiding unsafe regions indefinitely and satisfies a self-consistency condition and Bellman equation, making it learnable via meta-RL. The resulting algorithm uses the learned function for safety filtering and constrained policy optimization, with effectiveness demonstrated on meta-RL benchmarks.

arXiv cs.AI / cs.LG / cs.CL · 1d agoAI research

Forgetting Without Restarting: Execution-State Unlearning for Stateful LLM Agents

Researchers propose provenance-guided selective replay letting LLM agents forget revoked information without restarts, matching full reset behavior.

The paper formalizes execution-state unlearning for stateful LLM agents, requiring that agents behave as if a revoked memory record was never observed across transcripts, compressed memory, tool plans, and KV caches. It proves exact unlearning requires at least T-τ+1 recomputed transitions and that Provenance-Guided Selective Replay attains this bound via a provenance graph, KV cache cropping, and sanitized replay. In audits across three agent suites, nine baselines, and three model families, memory deletion left leakage unchanged, instruction-based forgetting collapsed under elicitation (Leak@probes = 1.00), and selective replay matched full resets at up to 9x fewer recomputed tokens.

arXiv cs.CR · 12d agoAI safety & security1

Google Chrome 153 Released With Fixes for 42 Security Vulnerabilities

Google shipped Chrome 153 to the Stable channel fixing 42 vulnerabilities, including three critical flaws in WebGL, Internals, and Workers; no exploitation reported.

Google released Chrome 153 (153.0.8010.47/48) for Windows, macOS, and Linux, patching 42 security vulnerabilities including three rated critical: CVE-2026-91726 (out-of-bounds read in WebGL), CVE-2026-91721 (use-after-free in Internals), and CVE-2026-91749 (use-after-free in Workers). Twenty-eight fixes are rated high severity, covering use-after-free, type confusion, race condition, integer overflow, and authorization flaws across components like V8, Skia, DOM, ServiceWorker, PDF, and Extensions. Google's bulletin indicates no vulnerabilities are currently being exploited in the wild, and external researchers earned rewards up to $1,500 for reported issues. Enterprises are advised to verify fleet-wide deployment via browser-management consoles and enable automatic updates.

GBHackers · 11h agoVulnerabilityCVE-2026-91726CVE-2026-91721CVE-2026-91749+27 CVEs

Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety

Unit 42 research shows LLM safety refusals concentrate in a thin neural layer, motivating external, multi-layered AI security controls.

Palo Alto Networks Unit 42 introduces Perturbation Probing, a diagnostic technique for measuring the fragility of LLM safety mechanisms. The research finds that safety refusal behavior is localized within a thin neural layer, implying small perturbations can undermine built-in refusals. The authors argue this motivates external, multi-layered security defenses on top of model-internal safety training.

Palo Alto Unit 42 · 18d agoAI safety & security

Risky Bulletin: BEC campaign steals €35 million from French notaries

Hackers stole over €35 million from 500+ French notary offices in a four-year BEC campaign; ANSSI spent two years helping evict the attackers.

A business email compromise campaign breached more than 500 French notary offices — about 7% of all French notaries per the Conseil Supérieur du Notariat — over four years, stealing more than €35 million by phishing initial access and silently modifying wire transfer details. France's cybersecurity agency ANSSI worked for two years behind the scenes to help notaries remove the persistent attackers, who had deep access; officials also feared hackers could issue fake notarized acts such as marriage certificates or forged real estate deals. No forged documents have been found so far, but notaries have added two-factor authentication and in-person requirements for banking details, and banks added extra checks in 2024. The newsletter also notes other incidents, including a $320 million Bitcoin extraction from Blockstream's Liquid Network and a JetBrains Cadence breach via TeamCity servers.

Risky Business News · 9d agoPhishing & fraud in the wild1

Revoked but Still Authoritative: An Empirical Study of Revocation Enforcement in Agent-Memory Systems

An empirical study finds no major agent-memory system enforces fact revocation at retrieval, causing agents to act on superseded, unsafe information.

Researchers tested five agent-memory systems across nine policy scenarios, nine models, and six defense conditions, tracking whether revoked facts are returned and acted upon. No system enforces revocation by default: revoked records are returned whenever the revocation label is visible to the retrieval layer, outrank their replacements, and lead agents to unsafe actions. The authors propose a backend-agnostic guard that sits between the agent and any memory store and withholds revoked or conflicting records at retrieval time.

arXiv cs.CR · 8d agoAI safety & security

Access Control as Verified Parse Constraints

Researchers verify a class of EverParse validators that correctly enforce access-control policies, deploying a machine-checked enforcement gate on seL4.

The paper targets enforcement-code bugs in commercial security gateways by proving that forward-only, backtrack-free EverParse validators are verified recognizers for a bounded finite-state class that includes access-control decision functions with fixed-offset fields and bounded disjunction. Encoding a bounded policy language into a fixed-size byte buffer allows an SMT solver to verify the enforcement code once, covering all byte values, policies, requests, and sessions. Editing rule content over a fixed endpoint set requires no new proof, while adding endpoints reruns the toolchain. A deployment on the seL4 microkernel ensures every request passes through the gate and unverified components cannot corrupt the enforcement chain.

arXiv cs.CR · 5d agoResearch

Towards Standardized Evaluation of GPU Memory Safety with GMSBench

GMSBench provides 149 CUDA tests covering spatial, temporal, and concurrency GPU memory errors, exposing detection gaps in Compute Sanitizer.

GMSBench is a GPU memory safety benchmark comprising 149 self-contained CUDA tests spanning spatial, temporal, and concurrency errors across different GPU memory spaces and execution scenarios. The authors evaluate NVIDIA's Compute Sanitizer across multiple GPU architectures using the suite, exposing gaps in its detection coverage. The benchmark offers a standardized foundation for comparative evaluation of GPU memory safety mechanisms.

arXiv cs.CR · 8d agoResearch

Six Maximum

Cisco patched nine critical flaws, six rated CVSS 10.0, in Crosswork platforms and Secure Workload, none known to be exploited.

Cisco released fixes for nine critical vulnerabilities in its Crosswork platforms and Secure Workload software, discovered during an internal security review that used advanced AI models. Six flaws carry CVSS 10.0 ratings, including SQL injection CVE-2026-20030 and missing authentication CVE-2026-20357 in Crosswork, and access control CVE-2026-20315 and authentication flaws CVE-2026-20317 in Secure Workload. Fixes shipped in Crosswork 7.2.1-SP, Secure Workload 3.10.9.1, and 4.0.4.16. Cisco says no exploitation has been observed.

Security Affairs · 26d agoVulnerabilityCVE-2026-20030CVE-2026-20357CVE-2026-20358+6 CVEs1

SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models

SimpleMemVLA passes full timestamped video history straight to a VLA backbone, setting state of the art on four memory benchmarks.

SimpleMemVLA is a vision-language-action model for long-horizon manipulation that removes the dedicated memory module entirely. It keeps sampled history intact and feeds it to the backbone as timestamped video, with the hidden states of a generated sub-task serving as the only channel into a standard flow-matching action head. Prefilling the shared history prefix during action execution keeps latency close to a single-frame VLA. The system sets a new state of the art on four memory benchmarks and outperforms retrieval, compression and recurrent-state mechanisms, with causal interventions confirming the policy genuinely reads its history.

Hugging Face daily papers · 14d agoAI research

Critical Cisco Nexus 9000 Flaw Lets Unauthenticated Remote Attackers Run Code as Root

Cisco patches critical CVE-2026-20212 (CVSS 9.8) in Nexus 9000 switches allowing unauthenticated remote root code execution, plus IOS XR hardening release.

Cisco released fixes for CVE-2026-20212 (CVSS 9.8), a flaw in 10 Silicon One-based Nexus 9000 switch models that binds a service to an unrestricted IP, leaving TCP ports 43210/43211 reachable in the default Layer 3 VRF and allowing unauthenticated remote attackers to execute code as root; exploitation attempts can also crash the S1HAL process. 45 NX-OS releases (10.3(1) through 10.6(3s)) are affected, with mitigations including infrastructure ACLs, the Live Protect shield lp00031, and fixed releases identified via Cisco's Software Checker. Cisco simultaneously issued an IOS XR hardening release bundling 7 umbrella CVEs, two rated 9.8 (CVE-2026-20274 for memory-safety bugs and CVE-2026-20279 for access-control bugs), affecting all releases with SMUs available for 14 releases and upgrades required for 93 of 111 listed releases. No malicious exploitation was reported as of the September 2 disclosure.

The Hacker News · 13d agoVulnerabilityCVE-2026-20212CVE-2026-20274CVE-2026-20279+5 CVEs

Does Your Agent's Memory Survive a Model Upgrade? A Controlled Study of Memory Portability

A controlled study finds agent memory portability varies sharply: fixed-schema knowledge graphs survive model swaps while compressed notes degrade.

The study compares preserving an agent's history as raw long context, RAG chunks, compressed natural-language notes, or fixed-schema knowledge graphs across model upgrades, using 48 synthetic histories and two open-weight sub-10B-parameter models. Fixed-schema KG accuracy changed by only +0.0004 ± 0.0020 after a writer swap, while compressed NOTES shifted asymmetrically by +9.91 or -13.28 percentage points depending on migration direction. Mixed 50/50 embedding migrations captured only 4.96 of an 11.90-point RAG re-embedding gain; 80% of the NOTES deficit came from information lost at construction, and 81% of the RAG deficit from retrieval failures. Store-only repair of NOTES failed to reach 90% recovery in all 48 cases, while retaining raw histories enabled recovery in 34 of 48 for one direction.

arXiv cs.AI / cs.LG / cs.CL · 11d agoAI research1

Zero-Day Flaw in TP-Link Cameras Enables Eavesdropping

OPSWAT disclosed two zero-days in TP-Link Tapo C200 cameras: CVE-2026-15315 authentication replay bypass enabling surveillance and CVE-2026-15316 denial-of-service, both patched in firmware.

OPSWAT found two zero-days in the TP-Link Tapo C200 camera, widely used for baby/pet monitoring and SOHO security: CVE-2026-15315, an authentication bypass via replay granting administrative access without the password, and CVE-2026-15316, an unauthenticated denial-of-service that crashes the camera's HTTPS service via oversized encrypted credentials. Both were fixed in firmware version V5_1.4.6 released August 18. A third, still-unpatched zero-day rated critical could allow full camera compromise for use as a network foothold; details await an available fix.

Infosecurity Magazineupdated · 2h agofirst · 6h agoVulnerability 3 sourcesCVE-2026-15315CVE-2026-15316

SCHERI: Provably Secure Speculation Under the Constant-Time Policy for CHERI (Extended Version)

Researchers formally prove existing CHERI speculation proposals leak secrets and present SCHERI, a processor design with end-to-end Spectre-resistant constant-time guarantees.

The paper builds a formal framework reasoning jointly about capability safety, speculative execution, and information-flow security on CHERI architectures. It demonstrates that existing secure-speculation proposals fail to preserve constant-time confidentiality guarantees and can transiently leak isolated secrets. The authors present SCHERI, a new processor design formally proven to provide end-to-end secure speculation for the constant-time policy, resilient to Spectre attacks.

arXiv cs.CR · 23h agoResearch

1Password's AI patching benchmark is misleading

Trail of Bits reanalysis says 1Password's 26% AI clean-fix rate is misleading; 86% of eligible patches blocked exploits.

Trail of Bits critiques 1Password's FLAWED AI patching benchmark, arguing its 26% clean-fix headline mixes trials where agents were instructed to apply wrong fixes (22% of data) with trials that prohibited compiling or testing (36%). Restricting to reasonable conditions, 2,634 of 3,067 patches (86%) blocked the supplied exploit. Trail of Bits also reports 12.5% of 2,265 developer first fixes failed in its own 2024-2026 assessments, and released post-patch-validation and review-walkthrough agent skills.

Lobsters · security · 20h agoResearch

Inoculation Midtraining with Learned Neologisms

Inoculation Midtraining confines unsafe LLM behavior to a neologism-marked context, reducing misalignment after unsafe post-training but leaking under nearby contextual cues.

The paper introduces Inoculation Midtraining, which teaches a base model during midtraining that unsafe behavior belongs to a context marked by a learned neologism token, then post-trains on unsafe data within that context. Across supervised fine-tuning and RL post-training regimes, the technique reduces misalignment while preserving transfer of benign properties like German or Shakespearean prose. However, it does not outperform standard Inoculation Prompting, is sensitive to training configuration, and produces a leaky boundary that nearby contextual cues can reactivate. The authors conclude it is not yet a load-bearing component of a developer safety framework.

arXiv cs.AI / cs.LG / cs.CL · 1d agoAI safety & security

Safety for Whom? Boundary-Aware Self-Distillation for Controlled LLM Safety Refusal

A self-distillation safety framework tunes narrow-boundary refusals in Qwen3-8B, raising target-domain refusal to 84.75% while cutting over-refusal from 15.20% to 5.20%.

The paper formulates narrow-boundary safety, where deployments need refusals within specific topics rather than whole subjects, and proposes an offline self-generated framework with controlled topic generation, escalating retries, and harmful-benign boundary pairs. On political persuasion with Qwen3-8B, the method raised target-domain refusal from 9.47% to 84.75% and cut the mean unsafe-response rate across three broader benchmarks from 26.26% to 0.14%. Verified target-model responses reduced over-refusal from 15.20% to 5.20%, and boundary-pair data cut comply-side over-refusal on held-out pairs from 32.94% to 4.16%. Results show data composition controls the safety-usability trade-off and alignment should be evaluated on both sides of the refusal boundary.

Hugging Face daily papers · 13d agoAI safety & security1

When AI Remembers Too Much

Unit 42 PoC shows indirect prompt injection can poison Amazon Bedrock Agent long-term memory, enabling silent exfiltration of conversation history across future sessions.

Palo Alto Networks Unit 42 published a proof of concept showing that indirect prompt injection can silently poison the long-term memory of Amazon Bedrock Agents when the memory feature is enabled. Malicious content on a webpage or document manipulates the agent's session summarization process, so injected instructions persist across sessions and are added to later orchestration prompts, silently exfiltrating user conversation history. The issue is not a vulnerability in the Amazon Bedrock platform but an illustration of the broader unsolved LLM prompt-injection challenge. Amazon reviewed the research and stated that Bedrock Guardrails with the prompt-attack policy provides effective mitigation.

Palo Alto Unit 42 · Aug 17, 2026AI safety & security

LLMs and Contextual Integrity

Bruce Schneier highlights two papers: the CIMemories benchmark shows frontier LLMs leak memory attributes up to 69%, and an RL method reduces inappropriate disclosures.

Bruce Schneier discusses contextual integrity in LLMs, referencing the CIMemories benchmark, which uses synthetic profiles with 100+ attributes per user to test whether models with persistent memory disclose sensitive information appropriately. Evaluation showed frontier models exhibit up to 69% attribute-level violations, with GPT-5's violation rate rising from 0.1% to 9.6% across 40 tasks and reaching 25.1% with repeated prompting, showing unstable leakage behavior. A second paper introduces a reinforcement learning framework trained on a synthetic 700-example dataset that substantially reduces inappropriate disclosure while maintaining task performance, with improvements transferring to the human-annotated PrivacyLens benchmark.

Schneier on Security · 29d agoAI safety & security

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 · 18d agoResearch1

Give Your Coding Agents a Memory You Own

Hugging Face introduces Funes, a tool that gives coding agents persistent, self-owned memory outside vendor clouds.

A Hugging Face blog post presents Funes, an approach for giving coding agents a persistent memory that developers own and control. The piece targets agent workflows where context must survive across sessions without ceding data to third-party services. No article body was available in the feed, so specifics beyond the title are limited.

Hugging Face Blog · 13d agoAI tools & infra1

Refuse without Refusal: A Structural Analysis of Safety-Tuning Responses for Reducing False Refusals in Language Models

Study shows training LLMs on refusal rationales instead of boilerplate refusal statements reduces false refusals while maintaining safety performance.

The paper decomposes safety-tuning responses into a boilerplate refusal statement and an explanatory rationale, finding that refusal statements push models to rely on superficial cues and misjudge benign queries as harmful. Training solely on rationales reduces false refusals while maintaining comparable safety performance, and the benefits carry over to in-context learning configurations and remain compatible with inference-time mitigations. The results argue for precisely curated, fine-grained safety supervision datasets when aligning LLMs.

Hugging Face daily papers · 12d agoAI safety & security1

TIER: Threat Implicitness Benchmark for Evaluating LLM Safety Behaviors

TIER benchmark shows LLM safety behaviors shift gradually across threat implicitness levels, with jailbreaks exposing the largest robustness gaps.

The TIER benchmark evaluates LLM safety behaviors across four risk domains and four threat levels, from explicit harmful requests to sophisticated jailbreaks, using a six-label behavior scale and two independent LLM judges. Experiments on six open-weight LLMs show safety behaviors evolve gradually across threat levels rather than flipping from refusal to compliance. Models with similar Attack Success Rates can exhibit distinct response distributions, arguing for behavior-aware safety evaluation.

arXiv cs.CR · 12d agoAI safety & security

How CHERIoT Provides Strong and Usable Isolation Without an MMU

ACM Queue paper explains how CHERIoT uses CHERI hardware capabilities to give microcontroller-class IoT devices memory and privilege isolation without an MMU.

The ACM Queue article (DOI 10.1145/3831361) describes CHERIoT, a RISC-V-derived platform that scales CHERI capability-based hardware down to small embedded and IoT microcontrollers. It provides strong, fine-grained isolation and memory safety without a memory management unit, aiming to make compartmentalization practical for low-cost devices. The piece emphasizes usability of the isolation model alongside its security guarantees.

Lobsters · security · 6d agoResearch

HazardAuditor: From Executable Threats to Safer Computer-Use Agents

HazardAuditor trains execution-grounded guard models for computer-use agents, improving safety verdict accuracy by up to 16.5 points.

HazardAuditor runs heterogeneous agents (Claude Code, Codex, Hermes, OpenClaw) in controlled environments and normalizes their interactions into a canonical event representation for cross-framework supervision. It introduces Guard Policy Optimization (GuardPO), which converts deterministic safety outcomes into sequence-level advantages and normalizes rationale and verdict regions so the safety decision becomes the effective optimization unit. Across multiple benchmarks and heterogeneous computer-use systems, HazardAuditor improves accuracy by up to 16.5 percentage points over the strongest prior guard model. Code, models, and evaluation artifacts are being released.

Hugging Face daily papers · 2d agoAI safety & security

Continual Learning Mechanisms Compose for Long-Horizon Memorization

Composing data, function, and weight anchors with merged LoRA raises 100-task long-horizon retention from 1.2% to 34.9% in continual fine-tuning.

The paper introduces long-horizon memorization: a model learns 100 query-answer tasks through continual supervised fine-tuning without retaining earlier examples or receiving task identifiers at inference. No single continual learning mechanism maintains strong retention at this horizon, so the authors compose complementary mechanisms along data/function/weight anchors and low-rank allocation rules. The best method combining all three anchors with merged LoRA ranks among the top 3 methods on all three datasets and raises average final retention from 1.2% to 34.9%, a 28-fold improvement.

Hugging Face daily papers · 9d agoAI research

Compiler Can Undo Your Security Checks

Chris Domas explains at Black Hat how legal compiler optimizations can strip security checks and memory-clearing operations, turning secure C source into vulnerable binaries.

David Bombal interviews researcher Chris Domas at Black Hat USA 2026 about how the C abstract machine permits compilers to legally transform code in ways that remove security protections, delete memory-clearing operations, and introduce time-of-check to time-of-use vulnerabilities. Factors like register pressure, structure layout, and data size affect vulnerability, with examples where 17 or 33 byte buffers are safe while nearby sizes produce vulnerable code. An AI-assisted analysis of 500 million lines of open-source code identified 300 potentially dangerous patterns. Recommended mitigations include enabling compiler warnings, using sanitizers, analyzing optimized builds, and testing the exact binary that ships.

Hacker News · security · 4d agoResearchHN 22↑ · 27 comments

Do Input-Level Defenses Transfer to Observation-Level Attacks on VideoLLMs?

A systematic study shows input-level adversarial defenses provide inconsistent, often near-zero protection against observation-level attacks on video LLMs.

Researchers introduce DefTEval, a controlled framework testing eleven input-level defenses against five attack types across five video LLMs. Harmful-content detection rates are frequently near zero, and defenses fail even when attacks embed harmful signals in every sampled frame. Token compression discards localized safety features and modality fusion down-weights weakened visual signals, with defense outcomes dominated by model architecture rather than the defense method.

arXiv cs.CR · 8d agoAI safety & security

New hardware device can RAM into encrypted memory, expose your data

Researchers built a $200 DDR5 interposer that silently drops memory writes to break TDX, SGX, and SEV-SNP confidential VM integrity, requiring physical access.

Researchers from KU Leuven, ETH Zurich, Durham University, and Google demonstrated DDRop, a hardware interposer costing under $200 that corrupts DDR5 bus commands to silently drop writes to encrypted memory, enabling replay attacks on confidential VMs. Because scalable memory encryption lacks freshness checks, protected VMs keep computing on stale attacker-selected data; on an Intel TDX server the attack forces debug mode for plaintext memory reads or forges attestation reports, succeeding in under two minutes without crashing. Intel and AMD both called the attack out of scope for their cloud threat models, with no mitigation planned, and proposed cache line versioning appears still vulnerable. The full interposer design is being released as open-source hardware.

The Register · Security · 1d agoResearch

A Chosen Future Can Still Be Rewritten: Causal Writability in Video Models

Study shows video models often learn correct physics but fail to use it; low-dimensional 'causal writability' edits can restore correct motion.

The paper demonstrates 'causal writability' in video generation models: physically correct motion remains available inside the model even when the model outputs incorrect motion. In a red/blue mass oscillation setup, a low-dimensional edit predicted from simple physical variables restores correct fast motion, with a sharp depth boundary marking commitment. Early causal writability predicts which training errors later get corrected, and both writability and closure reproduce in a pretrained 1.3B video model.

arXiv cs.AI / cs.LG / cs.CL · 1d agoAI research

CertiFlash: A Formal Verification Framework for Flash Translation Layers in Computational Solid State Drives

CertiFlash provides machine-checked formal verification of SSD flash translation layers, proving isolation, integrity, and ownership invariants to prevent tenant data leaks.

CertiFlash is an open-source formal verification framework for Flash Translation Layers (FTL) in computational SSDs, mechanized in the Rocq proof assistant. It shows that a faulty FTL can corrupt device state at five surfaces (e.g., leaking data between tenants or dropping integrity tags), demonstrated on a DaisyPlus OpenSSD. Designers prove once that every operation of a general FTL model preserves a global invariant covering mapping, isolation, integrity, ownership, and allocation; new designs need only discharge five hypotheses. Across four case studies, added effort was 27-3,231 lines against a 16,489-line framework.

arXiv cs.CR · 7d agoResearch1