Full Summary
This Wednesday morning, AI is dramatically accelerating cyberattacks, reducing the time from vulnerability disclosure to active exploitation to mere minutes. Both iTnews Asia and IT Pro report that AI automates the entire process, from monitoring disclosures to testing and exploiting weaknesses. Attackers are now scanning for new vulnerabilities within about 15 minutes of an announcement, often beginning exploitation attempts before security teams can even finish reading the advisory. This speed gap is forcing a rapid evolution in cybersecurity. The US Cybersecurity and Infrastructure Security Agency, CISA, now requires federal civilian agencies to fix critical security bugs in as little as three days, a significant reduction from previous 15 or 30-day mandates, as WIRED confirms. This urgency is driven by powerful AI tools like Anthropic's Claude Mythos and GPT-5.5, which, according to IT Pro, can discover and exploit software flaws at scale. The UK AI Security Institute has even shown these models successfully completing multi-step corporate attack chains. In response, companies are forging new partnerships and developing advanced AI security solutions. Samsung SDS is collaborating with XBOW and Tatum Security to enhance its AI-powered cloud security, while Enkrypt AI and Kode-1 are partnering to secure enterprise AI deployments across the APAC region. Databricks is launching new security features including a Private Network Gateway and automatic identity management. NetApp, Cisco, and Splunk have expanded their FlexPod AI solutions, automating incident response at the storage layer. Critically, AI is also being deployed to fight AI. BT has joined Anthropic's Project Glasswing, gaining access to the Claude Mythos Preview model, which has already found thousands of high-severity vulnerabilities. Synack and Wolfpack are bringing AI pentesting to South Africa, combining AI with human expertise for continuous security validation. Pi, an AI security platform, just raised $35 million to automate software protection and accelerate vulnerability remediation. However, challenges remain. CIO.com warns that companies are embedding AI everywhere without backup plans, leading to operational headaches during outages. Anthropic itself notes that releasing capable models like Claude Fable 5 comes with risks, requiring safeguards to prevent misuse. Even AI agents can be tricked: TechRadar reports an OpenClaw AI agent was exploited in phishing attacks after being granted sensitive access due to urgent-sounding requests. This means businesses need to plan for AI disruptions and increase the speed at which they apply security fixes, understanding that AI capabilities can become completely unavailable or even be turned against them.