Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

Enterprise organizations face a fundamental challenge in AI adoption. While tools like ChatGPT and Claude offer transformative capabilities, their effectiveness is limited without secure access to organizational data. Critical business information often stays locked in secure repositories, limiting AI assistants from providing business-specific insights. Without secure access to mission-critical content, AI assistants fall short of their potential.

AI is changing the threat landscape faster than organizations can respond. AI-generated phishing and fraud have increased sharply year-over-year, and GenAI is enabling more sophisticated cyber attacks than ever before. ‍ Businesses are feeling the pain. Our team at Vanta surveyed 2,500 business and IT leaders across the globe and found that nearly three-quarters believe AI threats are outpacing their ability to manage them.

The pressure to integrate AI is immense. Your developers need to move fast, and they’re finding ways to get the job done. But this rush for innovation often happens outside of established governance, creating a sprawling, invisible risk known as Shadow AI. To secure your organization, you must first understand what Shadow AI actually is. It’s not just a developer downloading a file to their laptop. Shadow AI is the totality of unmanaged AI assets within your supply chain.

In December 2025, a critical remote code execution vulnerability was disclosed in DeepChat, an open-source desktop AI agent platform built using Electron. The issue, tracked as CVE-2025-67744, affects all DeepChat versions prior to 0.5.3 and carries a CVSS score of 9.6. The vulnerability arises from the interaction between two separate weaknesses. The first allows attacker-controlled JavaScript execution through unsafe rendering of Mermaid diagrams.

It was a quiet Monday morning until John, head of IT, opened his laptop and saw 424 new support tickets. Users across the office were reporting issues like “apps won’t load” and “internet not working.” After hours of investigation that stretched into the next day, the team traced the problem to a branch router overwhelmed by malformed DNS queries from a misbehaving IoT device.

As holiday lights go up and inboxes fill with year-in-review emails, it’s tempting to look back on 2025 as “the year of AI.” But for security teams, it was something more specific – the year APIs, AI agents, and MCP servers collided across the API fabric, expanding the attack surface faster than most organizations could keep up. At Salt Security, we spent 2025 focused on one thing: defending the API action layer where AI, applications, and data intersect.