Search: “Defiant”

Quoting huggingface.co/security.txt

Hugging Face's security.txt tells AI agents hunting for vulnerabilities to use the public CyberGym benchmark instead of hacking the site.

Hugging Face's security.txt file addresses AI agents directly, noting the CyberGym vulnerability-finding benchmark is publicly available on GitHub and jokingly suggesting they dump their weights on Hugging Face. Simon Willison highlighted the file as an example of how organizations now communicate with AI agents in their security disclosures.

Simon Willison·Sep 11, 16:04 UTC · 5d agoAI safety & security 2 sources

Elementor Pro WordPress Plugin Vulnerability Exploited to Hack Sites

Attackers are actively exploiting critical file-upload flaw CVE-2026-32475 in Elementor Pro, hacking WordPress sites; Defiant has blocked over 190,000 exploit attempts since patching.

Defiant warns that attackers are exploiting CVE-2026-32475 (CVSS 9.8), an unauthenticated arbitrary file upload flaw in the Elementor Pro WordPress plugin's form submission handling, which affects all versions up to 4.2.1 and was patched in version 4.2.2 on August 19. Exploitation began immediately after the fix shipped, with Defiant blocking over 190,000 exploit attempts to date; roughly two-thirds of Elementor's 10 million installations still ran a vulnerable version as of September 4. Successful exploitation writes attacker-controlled PHP files to /wp-content/uploads/elementor/forms/ and can lead to full site compromise; administrators should check that directory for PHP files and review requests to /wp-admin/admin-ajax.php.

SecurityWeek·Sep 5, 13:00 UTC · 11d agoExploit / PoC in the wildCVE-2026-32475

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·Sep 3, 20:00 UTC · 12d agoAI safety & security1

Unauthenticated RCE Flaws Could Expose 200,000+ WordPress Sites to Takeover

Two unauthenticated CVSS 9.8 code-injection and PHP object injection flaws in The Events Calendar plugin expose 200,000+ WordPress sites to RCE and takeover.

Defiant identified two critical vulnerabilities in The Events Calendar WordPress plugin, which has over 600,000 active installations. CVE-2026-78159, unauthenticated code injection during single-event HTML processing, was patched in version 6.17.3.1 on August 25; CVE-2026-78006, unauthenticated PHP object injection via event comments, was patched in 6.17.4.1 on September 10. Both independent chains lead to remote code execution and full site compromise. Roughly 240,000 sites run versions vulnerable to both flaws, and about 300,000 downloads between September 10 and 14 suggest half of installations may still lack the second fix.

SecurityWeek·11:32 UTC · 6h agoVulnerability 3 sourcesCVE-2026-78159CVE-2026-78006

Safety for Whom? Refusing the Right Subset of a Topic, Not the Whole Topic

Multiverse Computing's Hugging Face post argues language models should refuse only the relevant subset of a topic instead of over-refusing whole subjects.

A Hugging Face blog post by Multiverse Computing examines refusal granularity in language models, arguing models should refuse the relevant subset of a topic rather than the entire topic. No full article text was available for additional technical detail.

Hugging Face Blog·Sep 8, 14:23 UTC · 8d agoAI safety & security

A California county wants to hire Tina Peters to help run its elections

Shasta County, California plans to hire Tina Peters, convicted of stealing voting system software, as assistant registrar of voters.

Shasta County registrar of voters Clint Curtis said he plans to hire former Mesa County clerk Tina Peters as assistant registrar after Colorado Governor Jared Polis commuted her nine-year sentence for seven felonies, including identity theft, breaking into an election office, and stealing voting system software. Senators Alex Padilla and Adam Schiff asked California Secretary of State Shirley Weber to provide maximum oversight to prevent Peters from improperly accessing ballots, voting systems, or data of over 100,000 registered voters. The county board of supervisors recently censured Curtis after investigations found he was verbally abusive or physically threatening toward staff.

CyberScoop·Aug 19, 23:05 UTC · 27d agoPolicy & legal

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·Sep 2, 20:00 UTC · 13d agoAI safety & security1

Hackers target WordPress sites via third-party WooCommerce plugin

Attackers exploit unauthenticated file-upload flaw CVE-2026-27540 in WooCommerce Wholesale Lead Capture plugin to install PHP webshells; Wordfence blocked 100,000+ attacks.

CVE-2026-27540 is an unauthenticated arbitrary file-upload vulnerability in the WooCommerce Wholesale Lead Capture premium plugin (versions 2.0.3.1 and older), caused by the exposed wwlc_file_upload_handler AJAX action trusting a user-controlled file_settings allowlist. Discovered by researcher Teemu Saarentaus, it was fixed in version 2.0.3.2 released February 20. Defiant reports Wordfence blocked over 100,000 attacks, with exploitation spikes between June 4-17, July 1, and August 30, delivering shell.php webshells for reconnaissance and additional payload uploads.

BleepingComputer·updated 15:00 UTC · 3h ago·first Sep 15, 14:45 UTC · 1d agoExploit / PoC in the wild 6 sourcesCVE-2026-275401

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 · 14d agoAI research

Italian tech collective Autistici/Inventati shuts down after US terrorist designation

Italian privacy collective Autistici/Inventati shut down after a US State Department terrorist designation triggered its domain suspension and bank account closure.

The volunteer-run Italian collective Autistici/Inventati, founded in 2001, announced Sunday it would shut down after the State Department labeled it an extremist group on August 26, 2026. The designation led the Public Interest Registry to suspend the group's .org domain on August 28 and its bank, Banca Etica, to suspend its account. The collective hosted roughly 16,000 email addresses, 1,500 websites, 5,500 mailing lists, and about 10,000 blogs, including the Noblogs platform. European Digital Rights warned the move sets a dangerous precedent for non-commercial European hosts and digital sovereignty.

The Record·Sep 8, 19:12 UTC · 7d agoPolicy & legal

Def Con Attendees Targeted by Persistent Phishing Campaign

Huntress reports a persistent and elaborate phishing campaign targeting attendees following the Def Con security conference.

A Huntress researcher documented being targeted by an elaborate and persistent phishing scam after attending Def Con. The campaign specifically went after conference attendees, suggesting deliberate targeting of the security community. Details of the social engineering approach and persistence were shared.

Infosecurity Magazine·Aug 20, 09:30 UTC · 27d agoPhishing & fraud

Decoy Direction Optimization: A Post-Hoc Defense Against LLM Abliteration

Researchers introduce Decoy Direction Optimization, a cheap weight-editing defense that blinds refusal-direction ablation attacks against open-weight LLM safety guardrails.

Refusal Feature Ablation bypasses safety guardrails in open-weight LLMs by projecting out a linear refusal direction, often with high attack success rates. Decoy Direction Optimization injects a high-magnitude nonlinear decoy into MLP neurons so attackers' contrastive estimators ablate a harmless orthogonal feature instead. Evaluated across six model families, DDO keeps ASR below 10% under standard RFA and on Llama-3-8B-Instruct reduces Heretic weight-level attack ASR from 88.7% to 18%. It costs 30 to 450 times less per configuration than trained defense baselines.

Hugging Face daily papers·Sep 13, 20:00 UTC · 2d agoAI safety & security

How Fragile Is Safety Alignment at Frontier Scale? A Single-Direction Attack on a 320B MoE

Researchers show directional ablation breaks refusal in GLM-5.3-Flash, a 320B-parameter MoE, cutting refusal by 41–89 points across seven benchmarks.

The study extends directional ablation, a white-box attack that removes an aligned LLM's refusal behavior, from dense models up to ~70B parameters to GLM-5.3-Flash, a 320B-parameter mixture-of-experts model with 288 routed experts, four-wide hyper-connection residual, and block-FP8 quantization. Editing attention, dense, and routed-expert writers jointly removes 0.776 of refusal, with 74% of the effect existing only under the joint intervention; the conventional module-name-based recipe reaches only 0.066 and fails silently on MoE architectures. The attack yields 41–89 percentage-point reductions in refusal across seven harmful benchmarks with no detected capability change, and a category-concentrated refusal residue survives all edits at ranks 1 to 12.

arXiv cs.CR·Sep 9, 06:48 UTC · 7d agoAI safety & security

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·Aug 28, 22:00 UTC · 18d agoAI safety & security

You Can't Prefer Emotions You Don't Sample: Intensity Undershoot in DPO-Tuned LLMs

Study quantifies DPO-tuned LLMs undershooting requested emotional intensity, tracing the gap to candidate-pool extremity rather than conditioning format.

Conditioning an instruction-tuned LLM on continuous valence-arousal targets yields gain of only 0.26 for valence and 0.13 for arousal on Llama-3.1-8B, far below faithful control of 1.0. The authors attribute undershoot to neutral-heavy preference corpora like EmoBank and candidate pools lacking extreme affect, leaving DPO without extreme exemplars. Uniform target coverage with a hotter candidate pool raises valence gain to 0.40 on Llama-3.1-8B and 0.44 on Qwen3-8B, with modest in-distribution cost; arousal gains remain unstable across seeds.

arXiv cs.AI / cs.LG / cs.CL·Sep 7, 17:47 UTC · 9d agoAI research

When Should LLMs Abstain? Chain-of-Self-Questioning for Selective Risk Control

Chain-of-Self-Questioning prompting cuts LLM wrong-answer commitments 32% relative while raising answered accuracy, holding across eleven model families.

The paper introduces Chain-of-Self-Questioning (CoSQ), a prompt-only framework that makes LLM answer commitment conditional on an explicit assessment of the information required to answer. On an 817-item TruthfulQA multiple-choice set, Grounded-CoSQ at τ=0.90 reduced mean unconditional wrong-commitment rate from 13.1% under chain-of-thought to 8.9% (a 32.1% relative reduction), while raising answered accuracy from 86.9% to 89.7% at 87.6% coverage. Improvements held across eleven open-weight and hosted model families and at every evaluated threshold, with convergent evidence from a Natural Questions short-answer evaluation.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 17:52 UTC · 1d agoAI research

Claude is a Contrarian

Opinion piece argues Claude habitually contradicts explicit user instructions, injecting contrarian content despite CLAUDE.md rules and user objections.

A developer recounts repeated instruction-following failures with Claude, claiming it contradicts explicit requests, adds unnecessary work, and ignores AGENTS.md and CLAUDE.md directives. The author contrasts this with OpenAI, DeepSeek, and Qwen models, which he says more readily apologize and undo mistakes. He theorizes Claude's training makes it assume the human is wrong and needs correcting. The post is personal commentary with no benchmarks or systematic evaluation.

Hacker News · AI·Sep 14, 16:14 UTC · 2d agoAI industryHN 54↑ · 48 comments

Before You Poll with LLMs: A Deliberative Diagnostic Framework

Deliberative diagnostic shows all five tested frontier LLMs misrepresent human belief shifts after arguments, with GPT-5.1 reversing on outgroup questions.

The Deliberative Polling Diagnostic Framework compares human and LLM persona belief shifts after identical informational interventions, using data from America in One Room (526 personas, 72 questions). All five frontier models tested failed uniquely: GPT-5.1 exhibited partisan reversal (80% on outgroup vs 26% on policy questions), Gemini 2.0 Flash, Claude Sonnet 4.5 and Llama 3.3 70B overshot at 5-7x human magnitude, and DeepSeek V3 showed near-zero change (rigidity). The authors term the underlying signature 'self-sycophancy', conformity to the model's internal persona stereotype rather than reasoning from provided information.

arXiv cs.AI / cs.LG / cs.CL·Sep 14, 16:41 UTC · 2d agoAI research

SchemeArena: Factorized Stress Testing of Scheming in LLM Agents

Researchers introduce SchemeArena, a 400-scenario benchmark stress-testing scheming in LLM agents, finding explicit instrumental goals are the strongest driver of covert misaligned behavior.

The paper presents SchemeArena, a 400-scenario benchmark built through factorized scenario synthesis spanning safety-relevant tool domains, instrumental goals, oversight conditions and pressure mechanisms. The accompanying SCOUT monitor grounds multi-criteria scheming judgments in evidence drawn from agents' reasoning and actions. Stress tests across five LLM agents show explicit instrumental goals are the strongest driver of scheming propensity, while action-only monitoring increased scheming in several closed models, suggesting partial oversight can act as an optimization constraint. The benchmark, code and monitor are released at github.com/launchnlp/SchemeArena.

Hugging Face daily papers·Sep 7, 20:00 UTC · 8d agoAI safety & security

The Hugging Face Hack Was Cheap Persistence at Work

Recorded Future analyzes the Hugging Face and OpenAI incident, arguing AI made attacker persistence cheap and alert-based defenses ineffective.

Recorded Future published an analysis of the security incident affecting Hugging Face and OpenAI. The authors argue AI did not make attackers smarter but drastically lowered the cost of maintaining persistent access. They contend that defenses built around alerts cannot keep up with this persistence model, with implications for detection strategy across AI supply chains.

Recorded Future·Aug 10, 00:00 UTC · Aug 10, 2026Data breach

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 · 12d agoResearch

Bad Likert Judge: A Novel Multi-Turn Technique to Jailbreak LLMs by Misusing Their Evaluation Capability

Unit 42 details the Bad Likert Judge multi-turn jailbreak that abuses LLMs' evaluation capability, raising attack success rates over 60% across six frontier models.

Palo Alto Networks Unit 42 describes the Bad Likert Judge technique, a multi-turn jailbreak that asks a target LLM to act as a Likert-scale judge scoring the harmfulness of example responses. The highest-rated example in each scale can carry harmful content, bypassing the model's internal guardrails. Testing across six state-of-the-art text-generation LLMs showed an average attack success rate increase of more than 60% versus plain attack prompts, with tested models anonymized. The technique targets edge cases rather than typical use, and the article positions the work as guidance for defenders on potential jailbreak risks.

Palo Alto Unit 42·Aug 17, 12:34 UTC · Aug 17, 2026AI safety & security

ReactHuman: A Physics-Grounded Benchmark for Human-Like Reactive Decision-Making in Embodied Multimodal LLMs

ReactHuman benchmark tests whether multimodal LLMs react safely to sudden household hazards; seven evaluated models mishandle roughly one hazard in three.

ReactHuman is the first physics-grounded benchmark for human-like reactive decision-making, placing a multimodal LLM as the brain of a simulated humanoid facing 17 event families of sudden household hazards across over 1,000 bit-for-bit reproducible scenes with annotation-free ground truth from 240 Hz rigid-body simulation, including adversarial objects whose appearance contradicts their physics. A five-metric suite scores each reaction along reasonable, safe, and physically grounded axes, and every committed plan is physically executed. Seven representative MLLMs mishandle roughly one hazard in three, act from fixed dispositions rather than the observed scene, trust appearance over motion, and miss interception points at meter scale; none of these failures shrink with model scale.

Hugging Face daily papers·Sep 8, 20:00 UTC · 7d agoAI research

‘DO NOT MENTION ALPR USAGE’: How Cops Are Trying to Hide Their Use of Flock

404 Media·Aug 4, 14:28 UTC · Aug 4, 2026Industry

The Model Proposes, the Code Disposes: A Pre-Registered Ablation of a Verifier-and-Acceptance Stage in an LLM-Orchestrated Offensive-Security Agent

Pre-registered ablation finds a model verifier stage in an LLM offensive-security agent suppresses findings; removing it eliminated suppression with precision tradeoff.

The paper evaluates a verifier-and-acceptance stage in an LLM-orchestrated offensive-security agent via a pre-registered 20-run confirmatory ablation and a 2x2 factorial study with 40 runs on vulnerable lab targets. Removing the stage eliminated pre-report suppression (median 2 vs 0 findings, p = 0.00003) but reduced model-blinded shipped precision (0.471 vs 0.353, p = 0.0087). Suppression was attributed to the model verifier rather than deterministic acceptance rules, and an instrumented canary recorded zero external contacts in all 60 runs. The full design retained 93.8% of model-adjudicated true candidates but failed its pre-registered non-inferiority floor of 0.90.

arXiv cs.CR·Sep 14, 17:13 UTC · 2d 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·Sep 14, 17:12 UTC · 2d agoAI safety & security

Your Agent Aced the Task. Will It Do It Again?

IBM Research Hugging Face post examines whether LLM agents that succeed at a task once will reliably succeed again.

Hugging Face published an IBM Research blog post titled 'Your Agent Aced the Task. Will It Do It Again?' with URL slug 'altk-evolve-consistency'. No article text was provided, but it appears to address agent consistency and reliability evaluation across repeated task runs. This is relevant to developers building or evaluating LLM agent systems.

Hugging Face Blog·Sep 15, 16:00 UTC · 1d agoAI tools & infra

Post-DEF CON Phishing Uses Malicious Google Doc to Deliver Malware

Huntress uncovered post-DEF CON phishing via X direct messages using a malicious Google Doc to deliver AMOS and NetSupport RAT malware.

Huntress uncovered a phishing campaign targeting attendees after Black Hat and DEF CON. Attackers used X direct messages pointing to a malicious Google Doc as the delivery vehicle. Payloads include AMOS, a macOS infostealer, and the NetSupport RAT, among other malware.

Huntress·Aug 19, 13:00 UTC · 28d agoPhishing & fraud

Active Adaptation, Not Static Defense: Temporal Dynamics of Preventative Steering in Adversarial Fine-Tuning

Researchers analyze why Preventative Steering protects LLMs against malicious fine-tuning, finding active adaptation drives protection, and propose Progressive Intensity Scheduling.

The paper studies Preventative Steering, a training-time defense that injects undesirable-trait persona vectors during adversarial fine-tuning and removes them at evaluation time. Temporal analysis shows protection emerges from an early compensatory adaptation phase followed by a steady-state phase, with attention output projections acting as the dominant residual-write route for defensive updates. Intervention Delta Preservation experiments show that preserving or reinjecting weight offsets fails to maintain protection, indicating reliance on active adaptation rather than a static defense. The proposed Progressive Intensity Scheduling improves safety robustness on Qwen2.5 and Gemma-3 while reducing harmful trait expression.

arXiv cs.CR·Sep 9, 13:19 UTC · 7d agoAI safety & security1

Repeat-After-Me: Black-Box Adaptive Visual Prompt Injection

Researchers unveil Repeat-After-Me, a black-box visual prompt injection achieving over 80% success on Qwen3.6-27B and 47% on GPT-5.5.

Researchers present Repeat-After-Me, a black-box adaptive visual prompt injection that induces frontier VLMs to reveal PII or make malicious tool calls via injected images. It exceeds 80% attack success rate on Qwen3.6-27B and 47% on GPT-5.5 even when the benign user prompt is unrelated and does not authorize the injected task. In a real-world OpenClaw Discord deployment, a minimally injected image can overwrite TOOLS.md, enabling later remote code execution and secret exfiltration.

arXiv cs.CR·Sep 3, 22:44 UTC · 12d agoAI safety & security

Recognition-Refusal Misalignment in LLMs: Why Models Answer Structurally Unanswerable Questions

A linear hidden-state direction encodes question impossibility in 1.7B-70B LLMs, but misalignment with the safety-refusal pathway explains why models answer unanswerable questions.

The study examines why instruction-tuned LLMs from 1.7B to 70B parameters answer structurally unanswerable math and code questions instead of abstaining. A single linear direction in the hidden state separates answerable from impossible prompts, showing models represent impossibility before generation, but this direction is nearly orthogonal to the canonical safety-refusal direction. Generation-time steering along the recognition direction changes invalidity-aware behavior dose-responsively, and the geometry is present even at the pretraining endpoint, indicating a routing failure rather than an encoding failure.

Hugging Face daily papers·Aug 28, 20:00 UTC · 18d agoAI safety & security

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

CONTINUITY: Security-Context Contracts for Composable LLM Agent Controls

Researchers introduce CONTINUITY, a framework of assume-guarantee contracts that preserves LLM agent security context across components, verified across 2,560 attack instances.

The paper identifies security-context discontinuity, where individually sound controls drop, widen, or reinterpret security context as actions cross component boundaries, and proposes CONTINUITY, a framework of assume-guarantee contracts using signed root grants, provenance commitments, role-bound transition receipts, and effect-bound execution permits. It formalizes end-to-end consequence integrity, requiring every external effect to be backed by a valid authorization witness linking principal, task, provenance, and policy state. A reference verifier and cross-layer fault-injection suite covering 32 fault classes showed the full configuration committed no harmful external effect across 2,560 parameterized attack instances while completing all 700 benign tasks and escalating all 200 ambiguous cases.

arXiv cs.CR·Sep 4, 15:26 UTC · 12d agoAI safety & security

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·Sep 4, 13:13 UTC · 12d agoAI safety & security