Uploaded April 2026 | Updated September 2026, 5 days ago
Since Anthropic introduced the Model Context Protocol (MCP) in late 2024, it has rapidly become a foundational standard for building AI agents that can securely call external tools and data. Thousands of start-ups are now building on top of MCP. Newsrooms, by comparison, have been slow to engage. This workshop argues that this hesitation matters. MCP is not simply another AI integration pattern. When combined with a fast-emerging ecosystem of open protocols for licensing, identity, access control, payments, and auditability, it offers media organizations a concrete path toward permissioned, paid, and measurable AI use of their content—something the open web has historically struggled to provide. The session will walk participants through a practical, newsroom-relevant technology stack aimed at newsroom leaders, product managers, and anyone thinking about journalism’s future distribution systems, including:
- Model Context Protocol (MCP) as the interface between AI agents and publisher systems
- Machine-readable licensing and access terms (e.g. Real Simple Licensing)
- Agent identity, authorization, and attestation, enabling publishers to know which agents are accessing which resources
- Wallet-less, usage-based micro-licensing and payments
- Audit trails and reporting for compliance, attribution, and revenue tracking Using real-world examples—including how organizations such as O’Reilly Media are enabling paid AI access—the speakers will demonstrate how these components work in practice. The workshop will also examine how some publishers are beginning to treat MCP endpoints as a new subscription surface for the agentic era, raising fundamental questions about distribution when content is accessed via personal AI assistants rather than websites or browsers. Finally, the session will address the risks. Despite new technical possibilities, Big Tech still controls the dominant personal agent interfaces (ChatGPT, Copilot, Gemini, Claude). Participants will explore what leverage publishers realistically have—and how protocol-level choices made today may shape news media’s negotiating position in the years ahead. Participants will leave with:
- A clear mental model of the emerging AI access and licensing stack
- Practical examples of how media organizations can begin experimenting now
- An understanding of both the opportunities and structural risks of an agent-mediated future The protocols are open. The question is whether journalism shows up.
Con: Florent Daudens (co-founder Mizal AI), Lucky Gunasekara (co-founder and CEO Miso.ai)
Since Anthropic introduced the Model Context Protocol (MCP) in late 2024, it has rapidly become a foundational standard for building AI agents that can securely call external tools and data. Thousands of start-ups are now building on top of MCP. Newsrooms, by comparison, have been slow to engage. This workshop argues that this hesitation matters. MCP is not simply another AI integration pattern. When combined with a fast-emerging ecosystem of open protocols for licensing, identity, access control, payments, and auditability, it offers media organizations a concrete path toward permissioned, paid, and measurable AI use of their content—something the open web has historically struggled to provide. The session will walk participants through a practical, newsroom-relevant technology stack aimed at newsroom leaders, product managers, and anyone thinking about journalism’s future distribution systems, including:
- Model Context Protocol (MCP) as the interface between AI agents and publisher systems
- Machine-readable licensing and access terms (e.g. Real Simple Licensing)
- Agent identity, authorization, and attestation, enabling publishers to know which agents are accessing which resources
- Wallet-less, usage-based micro-licensing and payments
- Audit trails and reporting for compliance, attribution, and revenue tracking Using real-world examples—including how organizations such as O’Reilly Media are enabling paid AI access—the speakers will demonstrate how these components work in practice. The workshop will also examine how some publishers are beginning to treat MCP endpoints as a new subscription surface for the agentic era, raising fundamental questions about distribution when content is accessed via personal AI assistants rather than websites or browsers. Finally, the session will address the risks. Despite new technical possibilities, Big Tech still controls the dominant personal agent interfaces (ChatGPT, Copilot, Gemini, Claude). Participants will explore what leverage publishers realistically have—and how protocol-level choices made today may shape news media’s negotiating position in the years ahead. Participants will leave with:
- A clear mental model of the emerging AI access and licensing stack
- Practical examples of how media organizations can begin experimenting now
- An understanding of both the opportunities and structural risks of an agent-mediated future The protocols are open. The question is whether journalism shows up.
Con: Florent Daudens (co-founder Mizal AI), Lucky Gunasekara (co-founder and CEO Miso.ai)
