Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

If your AI security strategy is focused on stopping copy-pastes into chatbots, you're fighting yesterday's war. AI agent security is now the real endpoint challenge — autonomous agents operating with your users' credentials. Traditional firewalls, CASBs, and DLP weren't built for agentic AI. They can't see delegated credentials, multi-step tool calls, or live runtime execution across IDEs, browsers, and local apps.

Autonomous AI agents execute tasks at machine speed, but they cannot take legal or operational accountability when code introduces security risks. Asaf Saar (EVP and Chief Product Officer, Mend.io) and Christian Jensen (VP Engineering, Tricentis) explain why human governance remains vital.

30,000+ devices compromised. 7,000+ crypto wallets targeted. Roughly $10M stolen. And that’s only one story. This week on The 443: Security Simplified, Marc Laliberte and Corey Nachreiner unpack a new wave of security incidents where identity, access, and AI collide. They cover: WaterPlum activity spanning 100+ countries More than 7,000 cryptocurrency wallets impacted, worth roughly $10M ShinyHunters’ claimed theft of 3 TB of FBI-related data An AI agent reportedly accessing Australian government health data beyond its intended permissions.

The barrier to entry for cybercrime is dropping fast. SPOILER: AI has a whole lot to do with it. Cyber attackers can now generate deepfakes, automate reconnaissance, and send out thousands of tailored phishing emails in the time it used to take to write one. Defenders are fighting back with AI of their own, catching threats earlier and shrinking the window attackers have to work with. Where do you land? Comment for attackers or for defenders and tell us why.