Security | Threat Detection | Cyberattacks | DevSecOps | Compliance

MCP rapidly enhances AI capabilities but introduces security challenges through its distributed architecture. Especially, the distributed nature of MCP requires a lot of NHIs and their secrets. Our research shows that MCP is a new source of leaks that already discloses real-world secrets.

In this post, we introduce our “Adversarial AI Explainability” research, a term we use to describe the intersection of AI explainability and adversarial attacks on Large Language Models (LLMs). Much like using an MRI to understand how a human brain might be fooled, we aim to decipher how LLMs can be manipulated.

It is no secret that AI is advancing at a blistering pace for everyone – including cyber adversaries. The mainstream unveiling of generative AI in 2024 brought about the biggest disruption in computing since the microprocessor. AI solutions promise the world unprecedented levels of efficiency, speed, and automation. However, this seemingly utopian and disruptive technology is evolving more quickly than governance and regulations can keep pace and implement guard rails for use and adoption.

The pace of innovation in generative AI has been nothing short of explosive. What began with users experimenting with public apps like ChatGPT has rapidly evolved into widespread enterprise adoption. AI features are now seamlessly embedded into everyday business tools, such as customer service platforms like Gladly, HR software like Lattice, and even social media networks like LinkedIn.

The best security outcomes come from the intersection of security expertise and the ability to act based on risk levels. At Arctic Wolf, we are laser focused on security outcomes for the security leaders and teams across our solutions — Arctic Wolf Managed Detection and Response (MDR), Aurora Endpoint Security, Arctic Wolf Managed Risk, Arctic Wolf Managed Security Awareness , Arctic Wolf Incident Response, as well as risk transfer with the Arctic Wolf Security Operations Warranty.

The pace and prevalence of adversarial AI is only accelerating — and organizations must harness the power of AI to stop them. CrowdStrike is building the tools they need to do it. Our latest innovations include Charlotte AI Agentic Workflows, Charlotte AI Agentic Response, and CrowdStrike Falcon Complete Next-Gen MDR now using CrowdStrike Charlotte AI. We are also announcing platform-level AI advancements that prioritize risk and aid in response.

Vanta is proud to be one of the first companies to achieve ISO 42001 compliance with our audit partner Schellman, an ANSI-accredited ISO 42001 auditor. ‍ To prepare for and pass our audit, our team worked diligently to assess our specific business needs, communicate clearly with stakeholders and AI leadership, and complete formal training to learn how to develop, integrate, and deploy trustworthy AI systems in line with emerging laws and policies.

Financial services can’t rely on manual review alone. Discover how unified data and explainable AI are helping firms detect risk, reduce cost, and stay ahead of evolving regulations. Financial services organizations are drowning in data. From emails and Bloomberg chats to WhatsApp messages and calls, the need to review communications data to detect potential misconduct and financial crime by employees and third parties is a mandated regulatory requirement for compliance and risk teams in 2025.