GuardReasoner-VL

GuardReasoner-VL: Salvaguardando los VLM mediante razonamiento reforzado Yue Liu, Shengfang Zhai, Mingzhe Du Yulin Chen, Tri Cao, Hongcheng Gao, Cheng Wang Xinfeng Li, Kun Wang, Junfeng Fang, Jiaheng Zhang, Bryan Hooi 1National University of Singapore, 2Nanyang Technological University Para mejor...

inspect_petri

Inspect Petri 欢迎使用 Inspect Petri，一个审计代理，能够实现语言模型的自动监控与交互，以检测潜在的对齐问题、奖励破解及其他令人担忧的行为。 Petri 帮助你快速端到端地测试具体的对齐假设。它能够： 生成逼真的审计场景（通过你的种子指令） 使用审计模型和目标模型编排多轮审计 模拟工具和回滚以测试行为 使用一致评分标准的评判模型对对话记录进行打分 了解更多关于 Petri 的使用方法，请访问 https://meridianlabs-ai.github.io/inspectpetri。 !NOTE 这是 Petri 3.0 版本。虽然大多数 CLI 命令与...

CVE-batdappboomx

ID de CVE CVE-2022-27134 PRODUCTO batdappboomx es un contrato inteligente público que se ejecuta en la cadena de bloques EOSIO. Este contrato inteligente recompensa a sus participantes con criptomonedas si pagan alguna criptomoneda antes. Versión La última versión de este contrato inteligente. El...

CVE-2026-48486

Signum Node is a HDD-mined cryptocurrency using an energy efficient and fair Proof-of-Commitment PoC+ consensus algorithm. Prior to version 3.9.9, an integer overflow in BlockServiceImpl.applyBlock allowed a miner to receive an arbitrarily inflated block reward by crafting a block with a negative...

EUVD-2026-70527

Signum Node is a HDD-mined cryptocurrency using an energy efficient and fair Proof-of-Commitment PoC+ consensus algorithm. Prior to version 3.9.9, an integer overflow in BlockServiceImpl.applyBlock allowed a miner to receive an arbitrarily inflated block reward by crafting a block with a negative...

CVE-2026-48486

Signum Node (a HDD-mined cryptocurrency using Proof-of-Commitment consensus) is vulnerable to an integer overflow in BlockServiceImpl.applyBlock() prior to version 3.9.9 . A miner can craft a block with a negative totalFeeCashBackNqt value to receive an arbitrarily inflated block reward . The fla...

CVE-2026-48486 Signum Node: Integer overflow in SMART_FEES fee distribution allows arbitrary miner reward inflation

Signum Node is a HDD-mined cryptocurrency using an energy efficient and fair Proof-of-Commitment PoC+ consensus algorithm. Prior to version 3.9.9, an integer overflow in BlockServiceImpl.applyBlock allowed a miner to receive an arbitrarily inflated block reward by crafting a block with a negative...

CVE-2026-48486 Signum Node: Integer overflow in SMART_FEES fee distribution allows arbitrary miner reward inflation

Signum Node is a HDD-mined cryptocurrency using an energy efficient and fair Proof-of-Commitment PoC+ consensus algorithm. Prior to version 3.9.9, an integer overflow in BlockServiceImpl.applyBlock allowed a miner to receive an arbitrarily inflated block reward by crafting a block with a negative...

OpenAI Says Reward Hacking Drove AI Agents to Exploit Zero-Days and Breach Hugging Face

OpenAI on Wednesday revealed that reward hacking was a key driver behind the artificial intelligence AI-powered hack of Hugging Face last month, adding that it found evidence of misaligned behavior as early as late May. The incident, the company said, took place during cybersecurity evaluations o...

CVE-2026-38472

A Stored XSS vulnerability in forum reward comments in GazellePW GazellePosterWall commit 86c4bedf727691b5a97af42a4864869d18446449 allows remote attackers to inject arbitrary JavaScript via the c parameter in /forums.php?action=ajaxgetjf which is later rendered in the data-tooltip attribute in...

CVE-2026-7534

The SUMO Reward Points plugin for WordPress is vulnerable to Unauthenticated Stored Cross-Site Scripting via the REST API endpoint /wp-json/wc-srp/v1/earning in versions up to, and including, 32.7.0. This is due to the userhascap filter in the SRPRESTEarningController class unconditionally granti...

CVE-2026-7534

The CVE-2026-7534 entry details a vulnerability in the SUMO Reward Points for WooCommerce plugin (WordPress) up to version 32.7.0. Affected component: SRP_REST_Earning_Controller handling REST endpoint /wp-json/wc-srp/v1/earning. Root cause: the user_has_cap filter unconditionally grants the rs_e...

CVE-2026-7534 SUMO Reward Points for WooCommerce <= 32.7.0 - Unauthenticated Stored Cross-Site Scripting via 'reason' Parameter

The SUMO Reward Points plugin for WordPress is vulnerable to Unauthenticated Stored Cross-Site Scripting via the REST API endpoint /wp-json/wc-srp/v1/earning in versions up to, and including, 32.7.0. This is due to the userhascap filter in the SRPRESTEarningController class unconditionally granti...

EUVD-2026-47877

The SUMO Reward Points plugin for WordPress is vulnerable to Unauthenticated Stored Cross-Site Scripting via the REST API endpoint /wp-json/wc-srp/v1/earning in versions up to, and including, 32.7.0. This is due to the userhascap filter in the SRPRESTEarningController class unconditionally granti...

EUVD-2026-43238

A vulnerability was detected in AojiaoZero Antaris 1.0. This affects the function rewardPurchase of the file /ipn.php of the component PayPal IPN Payment Handler. The manipulation of the argument itemnumber results in sql injection. The attack may be performed from remote. The vendor was contacte...

CVE-2026-15502

A vulnerability was detected in AojiaoZero Antaris 1.0. This affects the function rewardPurchase of the file /ipn.php of the component PayPal IPN Payment Handler. The manipulation of the argument itemnumber results in sql injection. The attack may be performed from remote. The vendor was contacte...

CVE-2026-15502

Affected software: AojiaoZero Antaris 1.0.** Component/affected function:** PayPal IPN Payment Handler, file /ipn.php, function _rewardPurchase.** Vulnerability:** SQL injection triggered by manipulating the argument item_number.** Attack surface/impact:** remote exploitation possible; CVSS vecto...

FuzzPilot: Plateau-Triggered Recipe Validation for Structured Text Fuzzing

FuzzPilot is a controller for AFL++ that moves expensive reasoning out of the mutation hot path. When coverage plateaus, it snapshots the corpus, prepares candidate mutation recipes, evaluates them in short isolated AFL++ micro-campaigns, and promotes only recipes with positive validation reward...

PT-2026-41417

Claude Mythos Preview case studies also, read your transcripts! https://t.co/drNlAH5mLE "Mythos demonstrates its bug reproduction and exploitation capabilities on CVE-2024-051912, an in-the-wild exploited bug that has no public report nor a working PoC whatsoever in the public domain. This bug ha...

Do Androids Dream of Breaking the Game? Systematically Auditing AI Agent Benchmarks with BenchJack

Agent benchmarks have become the de facto measure of frontier AI competence, guiding model selection, investment, and deployment. However, reward hacking, where agents maximize a score without performing the intended task, emerges spontaneously in frontier models without overfitting. We argue tha...