Microsoft’s AI Security Boost Microsoft introduced five partner-built AI agents through its Security Copilot platform. OneTrust’s Privacy Breach Response Agent automatically finds data leaks, maps affected countries, and creates reports for privacy teams. New phishing protection in Microsoft Teams blocks malicious links starting April 2025.

Securing AI Identities CyberArk and Accenture revealed plans to manage AI agent identities safely. Their system uses Accenture’s AI Refinery to check permissions and stop misuse. This helps organizations avoid attacks where hackers steal AI access keys.

Risky Bot Behavior A Delinea report showed non-human identities (NHIs)—like software bots—are often ignored. Over 70% of NHIs use old passwords, and 97% share access with outside companies. The Cloud Security Alliance found 1 in 5 companies had NHI-related breaches last year.

AI Phishing Surpasses Humans Hoxhunt’s tests proved AI creates better phishing emails than elite red teams. Since 2023, AI’s success rate jumped 55%, tricking even trained users. Behavior-based training remains the best defense.

New Tools & Rules Spot AI launched a universal agent builder letting businesses create custom AI guards for cameras using simple commands. Meanwhile, regulations now require detailed logs of AI agent decisions, especially in the EU. Audits ensure AI follows privacy laws when handling sensitive data.

Privacy-First AI Agents OneTrust’s agent automates breach responses, cutting work from days to hours. It flags affected regions, picks correct laws, and generates audit trails. Microsoft’s Entra tool also blocks unsafe AI app use, preventing data leaks into tools like ChatGPT.