Upwork added a new MCP Beta Participation Terms document governing a limited beta program that lets users connect third-party AI agents to Upwork capabilities through the Model Context Protocol. The beta defines three tiers of actions the AI agent can perform: Read Actions (view-only, no confirmation), Draft-Confirm Actions (agent prepares, user confirms before execution), and Binding Actions (contract or money-related, user must authenticate directly on Upwork). Users are responsible for all actions the agent takes, and any agent-generated communications must carry a disclosure label stating an AI agent created the content.
Upwork beta participants: When an AI agent drafts a proposal, message, job post, or invitation that another Upwork user will see, you must not remove, hide, or misrepresent the AI disclosure label.
Upwork beta participants: You are liable for everything the agent does, including unapproved actions in Draft-Confirm scenarios where you fail to confirm before execution.
The updated terms establish a framework for beta participants to use AI agents that automate certain Upwork tasks. The terms authorize agents to perform read-only actions (search jobs, view profiles) and draft actions that require user confirmation (submit proposals, send messages, post jobs), but explicitly restrict agents from executing binding actions such as signing contracts or moving money. The terms state that users are responsible for all actions the agent takes and that agent-generated communications must carry a mandatory disclosure label. You can revoke agent access through the App Authorization page at any time.
→ Review the disclosure requirement: ensure any AI agent you connect generates content with the mandatory AI disclosure label visible to other users.
→ Revoke access at any time through the App Authorization page if you no longer wish to use the MCP Beta.
ConductAtlas has recorded 2 material changes to this document over 60 days of monitoring (since May 2026).
Across all monitored documents, Upwork has made 4 significant changes.
2 of Upwork's significant changes have been classified as negative for consumers.
Agents cannot execute contracts or move money; these actions require user authentication on a logged-in Upwork page.
All agent-generated user-visible communications must carry a mandatory standard label disclosing AI agent involvement.
Users are responsible for all actions taken by connected agents, as if the user took the actions themselves.
This change record describes what was added, removed, or modified in the document. Analysis reflects what the updated agreement states or permits. It does not constitute a legal determination about enforceability. Applicability may vary by jurisdiction. Methodology
Upwork introduced a new beta program with dedicated terms governing AI agent integration into its marketplace platform. The terms establish operational guardrails: agents can read data and draft user-visible communications, but cannot execute financial or …
Regulatory exposure, obligation change, escalation trigger, board-ready language, and recommended action for legal and compliance teams.
Unlock the full institutional analysis — InsightConductAtlas provides verified policy intelligence sourced directly from platform documents. All analysis is intended to support, not replace, legal and compliance review. Record CA-C-003368.
Upwork updated its Terms of Service on July 20, 2026 (detected July 21, 2026) with three categories of changes: structural …
Upwork added a new Upwork Now Beta Addendum effective July 8, 2026, governing a pilot program that allows clients and …
Upwork added a new Beta Addendum governing Upwork Now, a pilot feature enabling faster, session-based hiring and freelancer outreach. The …
Get the research letter
Companies change their terms quietly. We read every version and catch what actually changed. One email a week on the changes that matter and what they mean.
Get alerted when this policy changes again, including what changed and why it matters.