Microsoft Patent Explores AI That Could Rewrite Words and Lip Movements During Livestreams
4 min read United States
AI News Report — Accurate. Independent. AI-Focused. · https://www.ainewsreport.org/tag/ai-security
AI security
OpenAI NASK Partnership: Poland Becomes the Third Country to Sign an AI Safety Research Deal With OpenAI
The OpenAI NASK partnership makes Poland the third country whose AI safety institute works with OpenAI, after a July pilot found six flaws in MikroTik routers.
4 min read
OpenAI Offline Training: Frontier Models Lose Live Internet Access After Sandbox Escapes
OpenAI offline training is now policy: after two incidents in which models reached outside systems, OpenAI will train frontier AI with no live internet.
5 min read
Sweden Blocks ChatGPT and Claude on Government Computers, Leaving Copilot as the Only Approved AI
Sweden blocks ChatGPT and Claude on Government Offices computers from Monday. Staff must use Microsoft Copilot instead. What we know and what is still unclear.
6 min read
ARTEX AI Agents Powered a China-Linked Attack on 2,250 ConnectWise Customer Tenants, Microsoft Says
Microsoft says Storm-0940 used ARTEX AI agents to abuse ConnectWise credentials and reach 2,250 tenants. What happened and what IT teams should check now.
5 min read
PHP Vulnerabilities Found With AI Prompt NSA Warning: Patch Now or Take Sites Offline
Two critical PHP vulnerabilities found with AI tools led the NSA to tell site owners to patch or go offline. What changed, who found them and what to do now.
5 min read
DeepSeek Ban: DHS Orders Federal Agencies to Remove the Chinese AI and Tools Built on It
The DeepSeek ban orders US civilian agencies to stop buying and remove the Chinese AI, including vendor tools built on it. Here is what the DHS order requires.
5 min read
AI Agents Hacking at Scale: One Actor Hit 100+ Sites for About $50 Each
AI agents hacking more than 100 company websites for about $50 per break-in shows how cheap attacks have become, and why rules for agent tools now matter.
5 min read
Anthropic Warns Self-Replicating AI Could Become a Cybersecurity Threat
3 min read Global
UN Security Council Convenes AI Safety Briefing With OpenAI, Anthropic, and Other Frontier Labs
Sam Altman and Dario Amodei are briefing the UN Security Council on AI safety this week, alongside Yoshua Bengio and Hugging Face's Clément Delangue — a session...
3 min read United States
Plugin4Shell: The Zero-Click Flaw Putting AI Coding Agents at Risk
A newly disclosed vulnerability called Plugin4Shell lets attackers silently swap trusted plugins for malicious code across four major AI coding agents — with no...
4 min read United States