OpenAI is adding support for Anthropic’s open-source Model Context Protocol (MCP) across its products. Company CEO Sam Altman announced on Wednesday that the Agents software development kit (SDK) is now offering MCP support, and several other of its products will soon adopt it as well. The protocol essentially standardises how AI systems connect to external data sources, including third-party repositories and data hubs, to ensure that the chatbots do not behave differently. Notably, MCP was introduced to the open community in November 2024.
OpenAI to Adopt MCP Across Its Products
In a post on X (formerly known as Twitter), Altman announced that the San Francisco-based AI firm will adopt the protocol, given its popularity among users and developers. While developers can find the support for MCP now in the Agents SDK, two more products will also adopt it in the coming days.
These two products are the ChatGPT desktop app and the Responses application programming interface (API). Interestingly, the company is reportedly working on a feature that will soon let its Team subscribers connect the chatbot to Google Drive and Slack. It is possible that when the feature rolls out, it will also support MCP. OpenAI said it would share more information in the coming months.
Last year, Anthropic open-sourced the protocol for the AI community. It solves a major challenge for AI systems. Most chatbots that are powered by large language models (LLMs) rely on internal databases to answer user queries.
However, there are times when an AI system is being used to answer queries based on the internal knowledge base. This would require the system to connect to third-party cloud servers, legacy systems, and data repositories. Since AI space is still in its nascent stage, the widely different systems have different methods and processes of how they connect to data hubs.
This leads to AI chatbots behaving differently when accessing these datasets. The issues range from latency problems to the file formats that are accessible to how the model processes the information. MCP solves it by offering a standardised protocol to connect to external data hubs.
Reacting to the announcement, Anthropic’s Chief Product Officer Mike Krieger said in a post, “Excited to see the MCP love spread to OpenAI! MCP has [become] a thriving open standard with thousands of integrations and growing. LLMs are most useful when connecting to the data you already have and software you already use.”
