Figma co-founder Dylan Field is saying the AI protocol everyone's betting on is broken, and then his company embraces it anyway. According to TechCrunch, Field aired his frustrations about MCP (model context protocol), the open-source standard designed to let AI agents tap into external data and tools, even as Figma announced 20 new MCP partnerships at its first Config conference held in India.
Field has criticized MCP for weeks. In late September, he publicly questioned Anthropic, the company that created the protocol, calling out the contradiction of a standard built for openness while Anthropic keeps its own resources locked behind closed doors.
"I find it interesting that Anthropic do not expose artifacts/other cloud resources created via MCP," Field posted on X, adding: "The debate is framed as open vs closed. Everyone always prefers open … obviously. And meanwhile issues of asymmetric context sharing are ignored."
When pressed directly by TechCrunch, Field said MCP "is not a perfect protocol" and has "tons of room for improvement." He also suggested the standard advantages "frontier labs" over everyone else, a pointed criticism of the AI giants shaping the rulebook.
Why is Figma moving forward with 20 new partners including Notion, Linear, ClickUp, and Mistral if the protocol is so flawed? Field explained the company wants to be open but worries about abuse. Figma's CEO clarified that exposing too much data through MCP opens the door to malicious misuse, and while most users are well-intentioned, the company must plan for worst-case scenarios.
Field is simultaneously criticizing the system and betting his platform on it. For Figma's user base, especially the designers and dev teams in India who rely on the tool daily, this signals the company wants to give access to user data, but on Figma's terms.
MCP is the new battlefield. Everyone claims commitment to open standards while protecting their own infrastructure. Field is willing to say so explicitly.
