Aggregator

2026年360数字安全生态合作伙伴渠道大会启幕

智能体正在让AI风险从“错误输出”转向“错误执行”

Versa has introduced a patent-pending zero trust architecture for the Model Context Protocol (MCP), applying zero trust principles to AI execution. The company said every AI-generated action is validated against user identity, role-based access controls, and system policies before execution, with human approval required when defined by administrators. The launch addresses a growing challenge as enterprises deploy agentic AI systems. A single prompt can trigger multiple actions across network and security environments, reducing visibility into … More →

The post Versa extends zero trust principles to AI agents and MCP workflows appeared first on Help Net Security.

Российский bug bounty впервые обогнал HackerOne: сумма превысила результат конкурента в три раза.

GitLab released GitLab 19.0 with expanded secrets management, agentic merge request workflows, improved CI pipeline visibility, support for self-hosted open-source models, and supply chain visibility enhancements. Engineering organizations shipping more code than ever are confronting the AI Paradox firsthand, as the surrounding workflows for securing credentials, reviewing and merging changes, enforcing pipeline standards, and running AI in regulated environments have not kept pace. GitLab 19.0 advances the platform’s agentic core by embedding those capabilities where … More →

The post GitLab 19.0 adds AI workflows, secrets management, and self-hosted model support appeared first on Help Net Security.

Total figure for fraudulent transactions Apple has blocked since 2020 now stands at over $11bn

Proton Pass, a secure, end-to-end encrypted password manager, added credential sharing through AI access tokens, allowing users to give AI agents access to selected items and monitor activity. To gain access, an agent must provide a reason for the request so users can see what actions are being performed. Access tokens are available with Pass Plus (included in Proton Unlimited), Pass Family, Pass Professional, and Proton Workspace plans. “AI access tokens are easy to set … More →

The post Proton Pass adds monitored credential sharing for AI agents appeared first on Help Net Security.

从 2026 年 4 月 30 日起，Meta 应沙特政府要求在沙特境内屏蔽了 NGO 组织 ALQST for Human Rights 和 Democratic Diwan，以及沙特研究员 Abdullah Alaoudh 和人权活动人士 Yahya Assiri 的 Facebook 账户。Meta 也应阿联酋要求地理封锁了一名学者的账号。自 2026 年 3 月以来，已有逾 100 个 Facebook 页面和 Instagram 账户受到了限制。沙特还要求 X 平台地理封锁知名沙特活动人士的账号，目前 X 尚未遵守该要求。

Глава Раменского объяснил срабатывание системы оповещения.

17万告警降至42条，Agent发现传统方式遗漏的隐藏入侵

5月19日至20日，2026年中国网络文明大会在广西南宁举行。本次大会以“文明网络空间  昂扬奋进力量”为主题，由中央网信办、中央精神文明建设办公室、中共广西壮族自治区委员会、广西壮族自治区人民政府共同主办。

近期，CNCERT监测发现银狐远控木马通过批量生成的高仿真钓鱼网站大规模传播，样本落地后将Shellcode注入系统关键进程执行远控，与境外C2服务器建立持久化连接，实现对主机的隐蔽控制。黑产团伙疑似利用AI工具大幅提升钓鱼页面制作效率……

近日，全国网络安全标准化技术委员会公布《人工智能应用伦理安全指引1.0》，标志着我国在人工智能应用治理领域迈出了重要一步。在当前生成式人工智能、智能体等新技术快速迭代、应用场景不断拓展的背景下，《指引》的制定旨在平衡技术创新与风险防范……

5月20日，2026年中国网络文明大会互联网企业社会责任分论坛在广西南宁举行，中国网络社会组织联合会正式发布《中国网络诚信发展报告（2026）》。

当前，人工智能国际会展已成为高流动性、强时效性、高价值数据密集交换的特殊数字物理混合场域，面临前所未有的数据安全挑战。因此，加强人工智能国际会展数据治理，具有重要现实意义。

网络国防与前沿模型结合已是大势所趋

Keepnet, an Extended Human Risk Management (xHRM) platform, today announced that its voice and SMS phishing simulation data contributed to the 2026 Verizon Data Breach Investigations Report (DBIR). The 2026 edition is the first to include voice and SMS phishing simulation data at this scale. The DBIR records this as “an increase of 40% in the median click rate” between phone-centric and email-based simulations (Verizon 2026 DBIR, p. 50). According to the report, phone-centric phishing … More →

The post Keepnet contributes voice and SMS phishing data to the 2026 Verizon DBIR appeared first on Help Net Security.

Как ИИ переписывает ценность работы багхантера.

关注高级攻防对抗技术热点，研究对手技术进行高级威胁模拟，研判攻击安全发展方向。