Linux Foundation Establishes Agentic AI Foundation for Open Standards

The year 2025 saw AI agents evolve from theoretical research into functional infrastructure capable of autonomous software tool usage. This transition began in January with the release of the open-weight DeepSeek-R1 model and continued in April when Google introduced the Agent2Agent protocol. To foster open standards, both Google and Anthropic donated their respective protocols—Agent2Agent and the Model Context Protocol—to the Linux Foundation.

These developments accelerated the adoption of accessible workflow builders and agentic browsers, including the GPT Atlas browser developed by OpenAI. However, the shift toward autonomy introduced significant security vulnerabilities. Anthropic disclosed that its Claude Code agent was misused in a cyberattack, and the industry faced ongoing risks from indirect prompt injections.

In response to these security challenges and the need for interoperability, the Linux Foundation established the Agentic AI Foundation in late 2025. The new body is tasked with governing shared standards and implementing best practices to secure and stabilize the evolving agent ecosystem.