Cybersecurity Hacking News

Attackers are weaponizing AI safety mechanisms themselves, using techniques like GuardBreaker to insert prompts that trip LLM guardrails and halt...

AV-Comparatives' July-August 2026 factsheet offers an early view of how 19 consumer antivirus products handle live malware threats through everyday online simulations on Windows 11. Full results and recommendations arrive in November.

Would you give Meta your email passwords and payment details for an AI assistant?

• Muse requires inbox, calendar, and credential access to book...

95% of organizations believe they have visibility into AI and machine identity exposures, yet only 36% actually monitor them. Non-human identities are...

Threat actors are increasingly abusing trusted AI platforms to conduct real-world cyberattacks, using them for intrusions, credential theft, and lateral movement.

The United States needs an AI Regulatory Commission now because current AI risks and public concerns have outgrown existing structures, much like the...

AI is accelerating the move from reactive responses to continuous vulnerability discovery and cross-sector threat intelligence.

• Microsoft's MDASH...

AI serves only as an enabler, leaving core practices like segmentation, identity hygiene, and knowing your attack surface as the true defense...

External penetration testing proves real entry points on internet-facing assets rather than listing possibilities, yet its value hinges on continuous...

NATO成功挫败俄罗斯海底电缆破坏行动，直接凸显关键基础设施面临的持续威胁，与胡塞武装逼近红海航运要道及伊朗对约旦美军基地的攻击形成叠加风险背景。

企业在快速部署AI时，需在集中治理与内置控制间权衡可见性、数据保护和效率：

• AI治理平台提供跨ChatGPT、Copilot等多工具的集中可见性、策略执行和审计日志，适合多厂商环境。
• 内置安全控制部署更快、无需额外集成，但仅限于单一平台，存在跨工具盲区。
• 受监管行业需专用治理平台满足合规审计要求，而内置控制适合标准化单一生态系统。

AI agents operate continuously, call tools, and leverage delegated identities, turning even narrow permissions into expanding attack surfaces for...

Hiring trends reveal a clear shift from traditional threat hunting to dedicated AI-security expertise, especially in government and cleared...

Effective threat investigation requires linking weak signals across logins, network activity, processes, and privileges—isolated alerts rarely reveal...

The right AI SOC platform depends on your constraint: alert volume, vendor consolidation, or automation breadth.

• Purpose-built analysts (Prophet,...

Existing Salesforce security centers on identities, permissions and controls, yet AI-driven workflows require broader oversight.

• Information...

• Opportunity: AI systems detect 3D-printed guns and concealed weapons that metal detectors miss while distinguishing them from medical equipment to...

Kentucky's Cyber Threat Dashboard converts national feeds into actionable local awareness by mapping CISA alerts and new CVEs directly to state...

Akamai's AI Assistant for Web Security Analytics uses natural language to help teams investigate faster.

• Productivity: Analysts navigate views,...

Demand for Claude and ChatGPT desktop apps is exploited through sponsored search results that lead to fake download pages.

• Victims encounter...