Linux Foundation launches Appia Foundation for AI safety standards

Linux Foundation Launches Appia Foundation and DocLang Working Group to Standardize AI Infrastructure

The Linux Foundation on Wednesday announced the Appia Foundation, a new initiative to create open specifications and conformity assessment frameworks for AI systems, adding to a broader push to bring standardization to the rapidly evolving AI ecosystem. The launch comes just over a week after the foundation unveiled a separate effort, the DocLang Specification Working Group, aimed at developing an AI-native document format.

Appia Foundation Targets AI Trust and Compliance

The Appia Foundation, hosted under the Joint Development Foundation, will develop modular specifications that bridge global AI standards with practical assessment tools, allowing organizations to demonstrate their AI systems meet safety, trust, and compliance requirements. Google, Microsoft, OpenAI, Arm, Mastercard, and Siemens are among the companies that have joined as members.prnewswire

According to the Linux Foundation, the framework is designed so that conformity evidence can be reused across the AI supply chain, potentially reducing duplicate assessments and compliance costs. The foundation’s specifications are organized across a “Requirements and Guidance” layer and an “Assessment Enablement” layer, providing testing criteria, evaluation guidelines, and component typologies to verify AI models, systems, and applications.slashdot

DocLang: A Machine-Readable Format Built for AI

Announced on June 9, the DocLang Specification Working Group was founded by IBM, Nvidia, and Red Hat, along with contributors ABBYY, HumanSignal, and Forgis. The group aims to develop an open, universal document format optimized for large language model tokenizers rather than human readers.prnewswire

Mark Collier, executive director of LF AI & Data, said the goal is to “develop a vendor-neutral, interoperable standard that helps organizations prepare document data for AI more reliably, transparently, and at scale”. DocLang uses a constrained XML vocabulary with a one-to-one mapping between document tokens and model tokens, preserving both semantic meaning and geometric layout. The format builds on DocLing, an existing open-source document processing toolkit already hosted by LF AI & Data.azalio

A Coordinated Standards Push

Both initiatives operate under vendor-neutral governance through the Joint Development Foundation. Together, they reflect the Linux Foundation’s expanding role as a convener for AI infrastructure standards at a time when governments worldwide are introducing AI regulations that demand verifiable compliance. The Appia Foundation addresses how AI systems are assessed against those regulations, while DocLang tackles how the document data feeding those systems is structured and exchanged.linuxfoundation