At RSA Conference 2025, agentic AI is at the forefront as a transformative force in cybersecurity. Unlike traditional AI copilots that assist users through prompts, agentic AI systems can autonomously execute complex, multistep tasks, making real-time decisions without human intervention. This shift is being propelled by the introduction and increased adoption of the Model Context Protocol (MCP), an open standard developed by Anthropic, which enables AI models to interact seamlessly with external tools, data sources, and APIs.

The Role of Model Context Protocol (MCP)

The widespread adoption of agentic AI is facilitated by the Model Context Protocol (MCP), which standardizes the way AI models access and interact with external data and tools. MCP allows AI systems to maintain context across various tasks and data sources, enabling more coherent and effective decision-making.

Anthropic's MCP has been embraced by major AI providers, including OpenAI and Google DeepMind, and integrated into platforms like AWS. This standardization simplifies the development of AI agents and accelerates their deployment across different environments.

Security Implications and Challenges

While agentic AI offers significant benefits, it also introduces new security challenges. The autonomy of AI agents necessitates robust security measures to prevent unauthorized actions and ensure compliance with policies. To address these concerns, companies like Bedrock Security are developing solutions like the MCP Server, which enables secure, context- and risk-aware agentic AI workflows. (​Source)

Additionally, researchers have proposed frameworks like MCP Guardian, which adds security layers such as authentication, rate-limiting, and logging to MCP-based AI systems, mitigating potential risks associated with autonomous operations. ​(Source)

Industry Adoption and Future Outlook

The integration of agentic AI is gaining momentum across various sectors. For example, Dataminr has launched Intel Agents to enhance real-time decision-making in its products, and Descope introduced the Agentic Identity Hub to simplify authentication for AI agents. (Source) 

As organizations continue to adopt agentic AI, the focus will likely shift toward establishing governance frameworks, ensuring ethical use, and developing standards for interoperability. The combination of agentic AI and protocols like MCP is poised to redefine cybersecurity strategies, offering more proactive and efficient defense mechanisms against evolving threats.​

These examples highlight how top cybersecurity and technology companies are leaning into Agentic AI – both to enhance their own offerings and to build new products geared to securing agentic AI. These trends are likely to be at the forefront of RSAC 2025.