Aggregator

When you ask a large language model to summarize a policy or write code, you probably assume it will behave safely. But what happens when someone tries to trick it into leaking data or generating harmful content? That question is driving a wave of research into AI guardrails, and a new open-source project called OpenGuardrails is taking a bold step in that direction. Created by Thomas Wang of OpenGuardrails.com and Haowen Li of The Hong … More →

The post OpenGuardrails: A new open-source model aims to make AI safer for real-world use appeared first on Help Net Security.

Bitdefender announced Bitdefender GravityZone Security Data Lake and Data Lake for Managed Detection and Response (MDR), solutions that help organizations cut through alert overload and complexity by unifying security telemetry from multiple tools into a single, intelligent platform. The new offering simplifies security operations with advanced analytics and expert-driven response, delivering the visibility of a modern Security Information and Event Management (SIEM) while reducing investigation time and total cost of ownership. Organizations face challenges with … More →

The post Bitdefender GravityZone Security Data Lake unifies telemetry from multiple tools appeared first on Help Net Security.

As LLMs, agents and Model Context Protocols (MCPs) reshape software architecture, API sprawl is creating major security blind spots. The 2025 GenAI Application Security Report reveals why continuous API discovery, testing and governance are now critical to protecting AI-driven applications from emerging semantic and prompt-based attacks.

The post Why API Security Will Drive AppSec in 2026 and Beyond  appeared first on Security Boulevard.

Prismatic announced its MCP flow server for production-ready AI integrations. The new offering enables companies to transform fragile AI toolchains into reliable, deterministic workflows that can securely power mission-critical applications. AI agents are now embedded in every workflow — from customer support to writing code to data analysis — but connecting them to real-world systems remains unreliable. Product teams expose individual API endpoints as MCP tools, leaving large language models (LLMs) to make dozens of … More →

The post Prismatic gives AI agents the guardrails and structure they need to operate reliably in production appeared first on Help Net Security.

11月4日，360数字安全集团主办的“纳米AI校园行”主题活动落地河南物流职业学院。活动聚焦“数字安全与人工智能技术讲座”“纳米AI视频创作赛作品征集”两大核心板块，旨在帮助学生深化数字安全与AI双重认知，推动课堂知识与产业需求精准对接，为河南省数字经济发展储备专业人才力量。

活动现场，360数字安全集团副总裁、ISC主理人卜思南率先带来前沿分享，围绕AI重塑生产生活场景以及应用落地焦点问题展开科普，并提出培育智能体工程师，用智能体来应对未来产业变革。此外，他深入探讨了AI从大模型到智能体，再到智能体蜂群的演进路径以及360针对AI安全风险探索出的“以模制模”创新路径。

分享中，卜思南用通俗语言拆解专业知识，不仅让学生直观感受到AI技术的魅力，也帮助大家清晰了解智能体的广阔应用场景。讲座尾声，他以“AI不是对手，是最强队友！武装自己，做AI时代的SUPERMAN！”鼓励学生，呼吁大家主动拥抱技术变革，借助智能体提升核心竞争力。

在当今人工智能技术飞速迭代的时代，人人学会运用AI工具、提升AI素养，不仅是适应数字时代发展的必然选择，更是推动个人成长与国家科技进步的关键所在。正是基于这样的时代背景与人才培养需求，360精心策划了“纳米AI校园行”主题活动，并同步推出“纳米AI视频创作赛”，旨在将国民级AI工具——纳米AI精准带入校园场景。通过这一举措让广大师生近距离接触前沿AI技术，在实践中提升对AI工具的应用能力，深化对数字安全与人工智能融合发展的认知，进而为数字时代下战略级人才的培养注入新动能，为国家数字经济的高质量发展储备专业力量。

关于“纳米AI视频创作赛”，纳米AI智能体产品专家韩煜东明确，赛事以“AI视界・影像中国”为主题，由360集团、华为云主办，新华网提供支持。作品评审阶段为11月20日至12月10日，公示阶段为12月11日至17日。大赛设置专业组（面向媒体机构、MCN机构、影视制作公司、自媒体创作者等）、院校组（面向全国大中专院校在校师生）、大众组（面向全体社会公众）三个组别，奖金金额从1000元至10万元不等，覆盖不同创作群体需求，成为ISC.AI学苑以赛事促实践的重要载体。

除了通过高规格赛事让学子接触前沿技术外，360还依托ISC.AI学苑这一SaaS化安全+ AI教育云平台深耕教育业务布局：一方面，整合自身数字安全与AI技术积淀，打造超4000课时的课程矩阵，为人才培育提供专业内容支撑；另一方面，针对高校 “师资实战能力不足、教学与产业脱节” 的核心痛点，通过现场实训、暑期研修、双师互聘等多元形式，助力打造区域 “双师型” 教师队伍。在此基础上，360还与全国多所院校深化合作，共建数智产业学院，将企业真实项目实训与本地化就业需求精准对接，推动教育链与产业链深度融合。通过这一系列举措，360定向培育兼具安全功底与AI创新能力的复合型战略人才，为筑牢国家数字安全屏障、抢占全球人工智能发展制高点夯实人才根基。

“360的教育布局始终以技术反哺教育，以教育赋能产业。”卜思南强调，ISC.AI学苑的核心价值在于将360的安全和AI技术沉淀转化为教育新质生产力。未来，360将持续推进 “纳米 AI校园行”，走进全国更多高校，让师生近距离体验纳米AI智能体等先进技术，进一步深化在校群体对AI应用的认知与实践体验。同时依托ISC.AI学苑平台形成支撑百万级师生的智能教育服务能力，将“河南模式”升级为全国可复制的“AI +教育”方案，持续扩大在 “教育AI+”领域的行业影响力。

Security teams are often surprised when they discover the range and number of devices connected to their networks. The total goes far beyond what appears in agent-based telemetry or old manual asset inventories. Enterprise networks face broader exposure from xIoT expansion Forescout analyzed 10 million devices across more than 700 organizations active in October 2025. About two-thirds of those devices weren’t traditional IT like workstations, laptops, servers, or hypervisors. Instead, they included network gear such … More →

The post Enterprises are losing track of the devices inside their networks appeared first on Help Net Security.

As AI investments surge toward $1 trillion by 2027, many organizations still see zero ROI due to hidden security and cost risks. Discover how aligning FinOps with security practices helps identify AI-related vulnerabilities, control cloud costs, and build sustainable, secure AI operations.

The post Using FinOps to Detect AI-Created Security Risks  appeared first on Security Boulevard.