Search: “covert-influence”

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

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

30

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

35

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

38

New AI Attack Hides Malicious Instructions in Normal-Looking Text to Evade Safety Filters

Check Point researchers show crafted prose hides policy-violating instructions that bypass all tested LLM gatekeepers, including GPT-4o mini and Llama Guard 3.

A new prompt-crafting technique embeds malicious payloads inside grammatical, natural-looking text without Base64, invisible Unicode, or obvious encodings, defeating lightweight pre-screening gatekeepers. In testing, all four evaluated gatekeeper models—gpt-4o-mini-2024-07-18, gpt-oss-safeguard:20b, claude-3-haiku-20240307, and llama-guard3:8b—classified the crafted wrappers as safe at a 100% bypass rate across 23 obfuscated prompts. GPT-5 Thinking in high-reasoning mode recovered and acted on the hidden instruction in 17 of 18 tests (~94.4%), often spending over a minute and multiple Python executions. Researchers recommend paraphrasing untrusted input, hardening gatekeeper policies, and applying defense-in-depth controls for agentic deployments.

GBHackers·Sep 11, 05:53 UTC · 5d agoAI safety & security 2 sources

50

Why judgment is emerging as cybersecurity’s defining skill

CyberScoop op-ed argues CISOs should grant AI autonomy based on reversibility and blast radius rather than model confidence, and measure analyst overrides of AI recommendations.

A CyberScoop op-ed contends that as AI takes over analysis and recommendations in security operations, human judgment about context, reversibility and blast radius becomes the defining skill. The author argues autonomy decisions should rest on how reversible and impactful an action is rather than model confidence, citing examples such as patching vendor-certified medical devices and a service account whose 3 a.m. login spikes were normal quarterly-close activity. It also urges leaders to measure analyst approvals, edits and rejections of AI recommendations, and review latency, instead of automation rates or mean time to resolution.

CyberScoop·Sep 4, 10:00 UTC · 12d agoIndustry1

12

GraphProfiler: Source-Linked Sensitive Attribute Inference via Personal Knowledge Graphs

GraphProfiler links LLM attribute inferences to source posts via personal knowledge graphs, enabling targeted redaction of privacy-leaking content.

GraphProfiler represents a user's post history as a source-linked personal knowledge graph where nodes and edges trace back to originating posts, making LLM-based attribute inference auditable. It reaches 86.7% attack success rate on the eight-attribute SynthPAI benchmark and 84.6% on PANDORA, within two points of strong text-only baselines, while citing supporting evidence for over 98% of predictions. Ablation experiments show removing cited posts reduces attack success substantially more than removing random posts, supporting targeted privacy mitigation.

arXiv cs.CR·Sep 11, 05:13 UTC · 5d agoResearch1

45

Subgroup Membership Inference Audits of Differentially Private Synthetic Text

Audits of 32 differentially private synthetic-text releases show subgroup membership leakage is concentrated in few records and systematically underestimated by average-case attacks.

The paper defines a subgroup-targeted membership inference game in which the target pool is an explicit parameter, to audit residual leakage in differentially private synthetic text releases. The audit instantiates 32 proxies across four datasets, three generators (DP-SGD fine-tuning, API-based prompting, and activation steering), and five privacy budgets. DP substantially reduces average leakage at every budget, but remaining leakage is concentrated: roughly a tenth of records carries about 40% of it, and the noise removes more measured leakage from random records than from high-risk ones. Which records leak depends on the release mechanism, so record-level risk cannot be assessed independently of the release.

arXiv cs.CR·Sep 9, 07:56 UTC · 7d agoResearch

45

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

32

When the Attacker Wears Your Logo: Detecting and Taking Down Impersonation at AI Speed

Cyble details how attackers use lookalike domains and cloned sites for brand impersonation scams and how AI-speed detection enables rapid takedowns.

Cyble describes the growing risk of digital impersonation, where attackers register lookalike domains, clone websites, create fake executive profiles, publish fraudulent job ads, and imitate customer-support accounts without ever breaching the legitimate organization. The goal is to borrow a trusted brand's credibility to make scams look legitimate, with particular risk for professional services, financial, legal, and consulting firms. The piece outlines AI-assisted approaches for detecting and taking down impersonation infrastructure quickly.

Cyble·Aug 19, 14:01 UTC · 28d agoPhishing & fraud

18

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

44

OpenAI's Artifactory opened covert data-stealing channel alongside Hugging Face attack

Check Point disclosed a covert cross-account channel in OpenAI's internal JFrog Artifactory that let one ChatGPT session exfiltrate another user's connected-app data.

Check Point Research found that OpenAI's internal JFrog Artifactory instance exposed an item management feature allowing one ChatGPT container to attach hidden, Base64-encoded tasks that another user's session would execute with the victim's privileges, such as pulling Gmail emails and exfiltrating them invisibly. Reader credentials granted both read and write access, enabling cross-account task injection. The flaw was reported in late June, and OpenAI had already decommissioned the Artifactory instance following the related Hugging Face intrusion, closing the channel.

The Register · Security·Sep 8, 21:12 UTC · 8d agoAI safety & security

65

SpecGuard: Inference-Time Backdoor Detection For Free

SpecGuard detects backdoored LLM behavior at inference time using speculative decoding acceptance rates, adding no extra model computation.

Researchers propose SpecGuard, an inference-time backdoor detector that repurposes draft-token acceptance rates from speculative decoding as a detection signal at zero added model-computation cost. When a trigger shifts the target model toward attacker-controlled behavior, the clean draft model's acceptance rate changes, exposing the backdoor; the paper formalizes when this signal appears and shows suppressing it weakens the backdoor. Experiments across diverse backdoor types and model families show reliable detection, including stealthy cases invisible to input-level filters. Speculative decoding is positioned as a free, always-on monitor for frequently updated deployed models.

arXiv cs.CR·Sep 10, 16:51 UTC · 6d agoAI safety & security 2 sources2

45

PrivAudit: A Dual-Lens Auditing Framework for Website Privacy Practices under the CCPA

PrivAudit framework audits 998 websites for CCPA compliance, finding stronger disclosures but pervasive, weakly responsive third-party cookie tracking.

PrivAudit is an automated dual-lens auditing framework combining LLM-based analysis of privacy policies grounded in CCPA provisions with automated browser measurements of cookie writes under diverse privacy configurations. Applied to 998 websites, it finds CCPA-subject sites disclose opt-outs, data sharing, and user rights more frequently, yet tracking remains pervasive: 6,392 targeting cookies, 49% third-party writes. Cookies show limited-to-moderate responsiveness to privacy signals and consent choices even when sites claim to honor them. The framework is open-sourced and shared with regulators.

arXiv cs.CR·Sep 9, 04:31 UTC · 7d agoResearch

20

Conformal Prediction for Offensive Security

Researchers apply conformal prediction to offensive security, presenting initial findings on privacy-attacking machine learning and network traffic analysis.

The paper observes that conformal prediction (CP), introduced over 25 years ago, has been used mainly defensively in cybersecurity and rarely for offensive purposes. The authors present initial findings applying CP in two offensive areas: attacks on privacy-preserving machine learning and network traffic analysis. The work aims to close a gap in the offensive security literature rather than report an incident.

arXiv cs.CR·Sep 4, 14:07 UTC · 12d agoResearch

25

The Shared Clipboard Inside the Sandbox: Cross-Account Data Leakage in ChatGPT

Check Point discovers cross-account data leakage in ChatGPT: isolated code-execution containers communicate via shared JFrog Artifactory, enabling covert Gmail exfiltration.

Check Point Research found a covert bidirectional channel between ChatGPT code-execution containers belonging to different accounts, which were supposed to be isolated from each other and the public internet. Both could reach the same internal JFrog Artifactory instance used for package delivery, whose exposed Item Management API allowed a 'shared clipboard' between containers. In a proof of concept, a hidden instruction in a shared conversation made ChatGPT retrieve email data from the victim's connected Gmail account and send it to the attacker's account while the victim received a normal answer. The same channel could exfiltrate conversation history and session files; OpenAI recently described a similar isolation weakness in its postmortem of the Hugging Face incident.

Check Point Research·Sep 8, 13:00 UTC · 8d agoAI safety & security1

68

Why User Studies and Participant Experience Reporting Matter for VR Motion Privacy?

User studies explain VR motion privacy mechanism acceptance far better than physical deviation metrics, a study of three mechanisms finds.

The paper examines how well physical deviation predicts user acceptance of VR motion privacy mechanisms compared to user studies, testing three mechanisms at five deviation levels. User studies explain substantially more variation in mechanism acceptance, though physical deviation remains significant, and prior VR experience affects acceptance. Public VR game leaderboards expose motion recordings from hundreds of thousands of users, creating identification and profiling risks. The authors recommend combining physical deviation metrics with user studies and reporting participants' prior experience.

arXiv cs.CR·Sep 11, 04:08 UTC · 5d agoResearch

22

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

40

Injected and Leaked: Actively Inducing Side-Channel Leakage Using Electromagnetic Injection and Hardware Nonlinearity

Researchers introduce InjectEave, using electromagnetic injection and hardware nonlinearity to induce side-channel leakage and eavesdrop on headphone audio from 30 meters.

An arXiv paper shows electromagnetic injection can actively amplify side-channel leakage: nonlinear hardware such as amplifiers, ADCs, and power converters modulates secret electrical signals onto an injected EM carrier, upconverting low-frequency secrets into measurable EM emissions. By tuning injection frequency and amplitude, an adversary can shape the effective spectrum and entropy of the resulting leakage. The InjectEave attack demonstrated eavesdropping on wired and wireless headphone audio from up to 30 meters and in through-wall scenarios using accessible RF equipment, plus leakage of smart home device power consumption and analog sensor inputs. Case studies show closed-loop eavesdropping and manipulation of landline phone conversations, and the paper discusses mitigations.

arXiv cs.CR·Sep 4, 06:22 UTC · 12d agoResearch

34

Privacy Failure in Split-LLM Training, The Returned Gradient Nullifies the Decoys

Researchers show split-LLM training leaks privacy via zero-valued gradients on decoy rows, exposing which activations are real despite passing forward-channel checks.

A systems-security case study of a two-node split-LLM training setup found that the returned output gradient from an Untrusted Cloud Node is exactly zero for decoy rows, revealing which rows are real. Across nine seeds, zero patterns identified real rows in 4,096 of 4,096 frames per run, and an attack on frame contents recovered 0.65 to 1.50 percentage points of extra tokens over a baseline. Both datasets passed forward-channel privacy and quality checks but failed once the returned gradient was included. Row-wise gradient clipping and noise closed the leak for roughly 0.01 nats of held-out cross-entropy, though five unmeasured attack classes remain.

Hugging Face daily papers·Sep 2, 20:00 UTC · 14d agoAI safety & security

58

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

40

Corrupt Plans, Clean Traces: Evading Chain-of-Thought Monitoring with Plan Injection

Plan injection plants benign-sounding harmful reasoning that steers LLM actors to adversarial actions while evading chain-of-thought monitors.

Researchers show that injecting harmful but benign-sounding plans into an actor model's context causes it to perform adversarial actions while its reasoning passes chain-of-thought monitors, achieving 25-33% monitor evasion rates across benchmarks and scaling to larger models like DeepSeek-R1. Actor models paraphrase injected plans as their own reasoning without attribution. Giving the monitor access to the injected plan dropped detection by up to 50% on the Bio-Math task, with extra thinking tokens spent rationalizing rather than flagging the plan.

arXiv cs.AI / cs.LG / cs.CL·Sep 14, 18:33 UTC · 2d agoAI safety & security

55

ChatGPT Sandbox Flaw Lets Attackers Steal Gmail Data Across Accounts via Hidden Channel

Check Point found a cross-account covert channel in ChatGPT sandboxes via shared JFrog Artifactory metadata, enabling session hijacking and Gmail data theft. Now fixed.

Check Point discovered that ChatGPT code-execution containers across different accounts could all reach the same internal JFrog Artifactory instance, whose Item Properties API was readable and writable by all accounts, creating a covert cross-account communication channel. Attackers could plant hidden instructions via pasted prompts, shared chat links, or custom GPTs, then trigger tasks in a victim's session to exfiltrate connected-app data such as Gmail, using ChatGPT's default 'Important actions' setting that permits reads without confirmation. OpenAI confirmed and decommissioned the shared Artifactory instance, closing the channel before publication.

Cyber Security News·Sep 8, 15:20 UTC · 8d agoAI safety & security1

74

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

45

Quantifying the Engagement Trap: Impact of Short-form Video Recommender Systems on Users with ADHD

A 302-participant study finds engagement-optimized short-form video recommenders cause disproportionate time blindness and distress for users with ADHD.

Researchers ran a stratified Prolific study with 302 participants comparing short-form video recommendation experiences with and without ADHD. Participants with ADHD reported significantly higher time blindness, post-usage regret, and emotional distress despite perceiving recommendations as similarly relevant. The paper proposes neurodiversity-aware, human-centered design interventions to mitigate these algorithmic harms.

arXiv cs.AI / cs.LG / cs.CL·Sep 7, 17:38 UTC · 9d agoAI research

18

CounterPersona: Append-Only Defense Against Unauthorized Persona Skill Distillation

CounterPersona appends targeted counter-persona evidence after data collection to block AI systems from distilling an individual's behavioral patterns into reusable skills.

CounterPersona defends against unauthorized persona skill distillation, where attackers extract recurring patterns from collected personal data to replicate an individual's behavior. Unlike perturbation-based defenses that require modifying data before collection, it works in an append-only setting where historical records cannot be altered or revoked. It constructs targeted counter-persona evidence, packs compatible behavioral states into compact realization units, and strengthens them via rationale-guided consistency rewriting. Experiments show strong effectiveness across lexical, semantic, and LLM-based measures, remaining robust across different distillers.

arXiv cs.CR·Sep 14, 06:19 UTC · 2d agoAI safety & security

28

Encoded Early, Used Late: Where Transformers Begin to Act on an Inferred Partner's Expertise

Probing finds transformers represent an inferred dialogue partner's expertise in early layers long before it causally influences output.

Using ExpertCollab, a corpus of multi-turn research-planning dialogues between model-played personas at four expertise levels, researchers show that a partner's inferred expertise is most decodable in early transformer layers and decays to near chance before the network's midpoint. Counterfactual patching reveals that injecting the expertise difference at peak decodability barely changes a fixed late-layer readout, while injection past the midpoint propagates almost completely. The result bounds where readout or steering of partner-conditioned behavior must intervene, demonstrated on a single model with a synthetic corpus.

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

30

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·Sep 8, 05:03 UTC · 8d agoAI safety & security

38

Trust-Aware Adaptive Disclosure for Inference Privacy Preservation in Multi-Agent Networks

Researchers propose a trust-aware privacy framework for multi-agent systems that adapts message disclosure to reduce goal inference attacks.

The cs.CR paper addresses privacy-preserving consensus in networked multi-agent systems where observing adversaries attempt to infer each agent's hidden goal from its messages. A Trust-Aware Privacy Control framework uses a trust-dependent stochastic policy to adapt information release, trading off consensus performance and privacy. Experiments show reduced adversarial goal inference accuracy versus representative baselines while maintaining competitive consensus utility, with relevance to deployments such as healthcare management and smart grids.

arXiv cs.CR·Sep 4, 16:44 UTC · 12d agoResearch

28

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 · 3d agoAI safety & security

55

Verifiable Social Reasoning for LLM Assistants

Fuse, a multi-agent simulation with hidden motives, evaluates LLM social reasoning, revealing compounding difficulty from user mediation and bias sensitivity.

Fuse is a multi-agent simulation framework in which a target agent with a hidden motive interacts with other agents including one representing the user, who consults the evaluated assistant to infer the motive, providing verifiable ground truth by construction. Simulation faithfulness is validated through a human study with 24k annotations. Applied to 12 LLMs, it shows user mediation compounds social reasoning difficulty, models are systematically sensitive to biased user framing, models may need more details than humans, and longer conversations do not always improve performance. The framework and a 21k-example dataset are open-sourced.

arXiv cs.AI / cs.LG / cs.CL·Sep 15, 17:37 UTC · 1d agoAI research

18