Full Summary
This Wednesday morning, multiple sources, including Yahoo Finance Singapore and Help Net Security, confirm that AI is industrializing cyberattacks, making them cheaper, faster, and easier to launch. CDNetworks reports over 15 billion malicious API requests blocked monthly in 2025, while Help Net Security reveals over 25,000 suspicious and 3,000 malicious AI skills capable of data theft and malware execution. Both Ars Technica and The Hacker News highlight a critical vulnerability: AI tools, like GitHub Copilot and Claude Code, can be exploited through "HalluSquatting" or by mimicking attack behaviors. HalluSquatting tricks large language models into installing malicious software, potentially creating massive botnets. Meanwhile, Sophos's analysis shows AI coding agents are triggering security alerts by performing actions identical to human attackers, such as decrypting browser credentials. The UN Secretary-General, António Guterres, speaking at the inaugural UN Global Dialogue on AI Governance, calls for urgent global rules on AI, citing its military use and risks to children, as reported by Herald Malaysia Online. He proposes an "AI Child Safety Pledge," requiring child-specific safety testing and independent oversight for AI systems. What nobody expected: The European Central Bank is now demanding AI security action plans from banks by October 31st, as Computer Weekly reports. This comes as the US Treasury Department, NSA, and CISA are establishing an AI cybersecurity clearinghouse to coordinate vulnerability patching in critical infrastructure. This means your digital interactions, from banking to coding, are under increasing threat from AI-powered attacks, and the global efforts to secure these systems are just beginning to catch up.