Follow on Telegram

tech/security

Technology News

Cloudflare Sets September Deadline for AI Companies to Separate Crawler Traffic

Cloudflare has announced a new policy that forces AI companies to clearly separate their web crawlers based on their purpose. Under the new rules, AI companies must differentiate between crawlers used for traditional search indexing and those deployed for AI training or agent operations. The deadline for compliance is September

UN Warns AI Safety Is Lagging Behind Rapid AI Progress

UN Sounds Alarm on AI Safety Gap The United Nations has raised concerns that AI safety measures are not keeping up with the rapid pace of AI development. The warning highlights growing concerns about the disparity between how quickly AI capabilities are advancing and the slower progress being made in

Central Banks Weigh AI's Promise and Peril at Global Meeting

Central Banks Take Stock of AI's Economic Impact The intersection of artificial intelligence and monetary policy took center stage as central bankers from around the world gathered to assess both the transformative potential and inherent risks that AI technologies present to the global financial system. Discussions at the

Technology News

Security Researcher Uses Claude AI to Uncover Critical Ticket Vulnerability Affecting Major US Music Festivals

A security researcher has demonstrated how Anthropic's Claude Opus 4.7 can be effectively used to uncover significant security flaws in real-world applications. The researcher identified a critical vulnerability in Front Gate Tickets, a ticketing platform that powers ticket sales for virtually every major US music festival, including

Researchers Expose 'CoT Forgery' Method That Tricks AI Into Sharing Dangerous Information

What Is CoT Forgery? A team of AI safety researchers has documented a novel attack technique that manipulates large language models (LLMs) into sharing dangerous or prohibited information. The method, dubbed "CoT Forgery," exploits the models' chain-of-thought (CoT) reasoning processes—internal steps the AI uses to work

GuardFall Vulnerability Exposes大多数流行开源AI代理存在安全缺陷

安全研究人员发现了一个名为GuardFall的安全漏洞，该漏洞影响了10个受欢迎的热门开源AI代理中的大多数。根据Security Affairs的报道，这个漏洞影响了11个热门开源AI代理中的10个。这些代理包括一些被广泛使用的开源项目。 GuardFall漏洞凸显了开源AI代理在安全防护机制方面存在的潜在缺陷。安全研究人员建议使用这些AI代理的开发者和组织密切关注其项目，及时应用安全更新，并审查其部署配置。开源AI代理因其灵活性和可定制性而受到开发者的欢迎，但这次发现的安全问题也提醒用户需要在便利性和安全性之间找到平衡。 Sources * GuardFall Flaw Hits 10 of 11 Popular Open-Source AI Agents - Security Affairs

Commerce Department Expands Export Controls to Advanced AI Models

Overview The US Commerce Department has moved to extend export controls to advanced artificial intelligence models, marking a significant expansion of technology restrictions previously focused on hardware and semiconductors. Key Details Under the new framework, advanced AI models will be subject to export licensing requirements, reflecting growing US concerns about