\n\n
Key developments: 1) Anthropic launched a free, opt‑in AI vulnerability scanner (OSS Scanner) that uses its Claude models to periodically scan open‑source projects, a move aimed at shoring up supply‑chain security. 2) Microsoft’s latest report warns that AI is automating ransomware and other attacks, stressing the need for stronger identity and data‑governance controls. 3) Researchers identified a surge in AI‑driven phishing, with fake ChatGPT, Gemini and Claude ad portals siphoning credentials and MFA codes. 4) Anthropic also expanded its Cyber Verification Program, giving vetted security teams reduced‑safeguard access to its models, after Project Glasswing uncovered over 129,000 software flaws. 5) New cybersecurity‑focused LLMs such as OrcaCyber Zero 1.5 are achieving near‑perfect benchmark scores, signalling a wave of AI‑assisted vulnerability research.
Why it matters: These AI tools dramatically increase the speed at which both attackers and defenders can discover and exploit weaknesses. Automated scanning and AI‑crafted phishing lower the skill barrier for threat actors, while the same technology offers defenders unprecedented ability to find and patch flaws quickly. The race between AI‑enhanced offense and defense heightens the urgency for phishing‑resistant authentication, zero‑trust identity models, and rigorous governance of AI agents.
What to watch next: • Adoption of AI‑driven vulnerability scanners by major open‑source ecosystems. • Further disclosures of AI‑generated phishing campaigns targeting enterprise credential flows. • Evolution of regulatory guidance on AI agent governance and zero‑trust frameworks. • Performance and accessibility of next‑generation cyber LLMs like OrcaCyber Zero 1.5 and their impact on vulnerability remediation pipelines. [[1, 2]]