AI Coding Agents for Enterprise: IP Indemnity, Data Residency and 500-Seat Cost Compared

Our ‘Top AI Coding Agents and Development Platforms‘ guide covered what each AI coding agent does and where it fits. This piece is for a different reader. It is written for the procurement lead, the general counsel and the security reviewer.

Those readers ask 4 questions before any rollout. Who pays if generated code triggers an IP claim? Where do prompts live? What can admins log? What do 500 seats actually cost?

We read the current contract language for GitHub Copilot, AWS Kiro, Cursor, Devin and Windsurf. Every term below was checked against the vendor’s own pages on September 26, 2026. This is reporting, not legal advice. Run final terms past counsel.

First, 5 Brands are Really 4 Contracts

Cognition acquired Windsurf. On June 2, 2026, Cognition renamed the Windsurf editor Devin Desktop. Devin and Windsurf now sit on one pricing table and one set of terms.

The Indemnity Table Nobody Publishes

This is the clause that blocks enterprise rollouts. Each row uses the vendor’s current contract. Quotes are short and taken directly from the linked source.

Axis 1: Legal Exposure on Generated Code

GitHub Copilot: Microsoft’s April 3, 2026 change matters more than any feature launch. Before it, one admin setting could void coverage. Microsoft’s required-mitigations page now lists no additional mitigations for GitHub offerings. The remaining limit is the word “unmodified”. Most shipped code gets edited, so ask counsel how far edits move code outside scope.

AWS Kiro: Kiro inherits AWS’s generative AI indemnity. AWS describes it as uncapped for copyright claims on output from services listed in Section 50.10. Coverage drops if your inputs infringe or you disable filtering features. Kiro states that paid subscribers ship with this protection included.

Cursor: The MSA is the broadest on paper. It names Suggestions and removes indemnity from the fee cap. The exclusions still bite when a claim relates to a modification or an unapproved combination. The consumer Terms of Service run the other way: users indemnify Anysphere.

Cognition (Devin, Windsurf): Cognition is the outlier. Its MSA defines Customer Data to include Outputs, then excludes Customer Data from indemnity. Generated code therefore sits outside the standard promise. Indemnity is also capped at 2x trailing 12-month fees. Any Devin or Devin Desktop rollout needs this fixed in the order form.

Axis 2: Where Prompts Are Stored

GitHub Copilot: For Business and Enterprise, the plans FAQ says IDE chat and completion prompts are not retained. Prompts from other surfaces, such as github.com, mobile and CLI, are retained for 28 days. User engagement data is kept for 2 years. GitHub does not train on Business or Enterprise data.

Enterprises on GitHub Enterprise Cloud with data residency can pin Copilot inference to the United States or the European Union. That policy adds 10% to AI credit consumption. Each region also exposes a narrower model list.

AWS Kiro: Per Kiro’s data protection page, enterprise content is never used for service improvement. It is stored in the region where your Kiro profile is configured. Inference stays inside the US or Europe geography, except for models tagged experimental. Admins can encrypt data with customer-managed KMS keys.

Cursor: With Privacy Mode on, Cursor holds zero data retention agreements with all model providers and does not train on your code. File contents are cached temporarily, encrypted with client-generated keys. Cursor’s public pages do not offer a customer-selectable processing region. Every request still passes through Cursor’s backend.

Cognition: The MSA bars training on Customer Data without written consent. The Customer Dedicated Deployment runs Devin’s devbox in a single-tenant VPC linked by AWS PrivateLink. The agent’s reasoning layer still runs in Cognition’s cloud. On self-serve paid tiers, Cognition may train on Customer Data until the customer opts out. On Teams, only an admin can opt out.

Axis 3: Audit Logging and Admin Controls

GitHub Copilot: The enterprise audit log records Copilot events under action:copilot. Agent activity appears under actor:Copilot for 180 days, per GitHub’s agent audit reference. GitHub is explicit that prompts sent locally are not in the audit log. Admins also set model, preview-feature and spending policies.

AWS Kiro: SSO runs through IAM Identity Center, Okta or Entra ID. Admins can enable prompt logging and daily user activity reports. Both land in an S3 bucket inside the customer’s AWS account, so prompt records stay under your control.

Cursor: Teams includes SAML or OIDC SSO, team-wide Privacy Mode and usage analytics. Audit logs, SCIM, repository and model access controls, and an AI code tracking API require Enterprise.

Cognition: Teams offers an admin dashboard with analytics. SAML or OIDC SSO, centralized admin controls and VPC deployment are Enterprise only. Cognition’s Enterprise API exposes audit log endpoints.

Axis 4: True Per-Seat Cost at 500 Seats

List prices below are monthly, in USD, before usage overages and taxes. The ‘true’ figure is the cheapest tier that delivers indemnity plus SSO.

Sources: GitHub Copilot billing docs, GitHub pricing (Enterprise Cloud from $21 per user), Kiro pricing, Cursor pricing, Devin pricing.

Two meters sit on top of every seat. GitHub moved Copilot to usage-based AI credits on June 1, 2026, where 1 credit equals $0.01. Kiro bills extra credits at $0.04, and enterprise overages are off by default. Agent-heavy teams should model usage, not just seats.

Feature Comparison: Copilot vs Kiro vs Cursor vs Cognition

Interactive Explainer: Run the Contract Scenarios

Toggle a claim scenario, trace where a prompt travels, and price your own seat count.

What to Put in the RFP

• Ask each vendor to confirm in writing whether edited or agent-written code stays inside indemnity scope.
• For Cognition, redline the Customer Data exclusion so Outputs are covered, and lift the 2x supercap.
• For Copilot, confirm whether you buy under GitHub terms or Microsoft Product Terms. The governing document differs.
• Require region commitments in the order form, not just a settings page.
• Price 12 months of agent usage at your real model mix, not the seat line alone.

Key Takeaways

• Copilot offers uncapped indemnity for unmodified output, with no filter requirement since April 3, 2026.
• Kiro is the cheapest 500-seat path with indemnity, SSO and customer-owned prompt logs.
• Cursor’s MSA names Suggestions and removes indemnity from the fee cap.
• Cognition’s standard terms exclude Outputs from indemnity, covering Devin and Windsurf alike.
• At 500 seats, Cursor and Cognition both require custom Enterprise pricing.

FAQ

• Does GitHub Copilot still require the duplicate detection filter for IP indemnity?No. Microsoft’s required-mitigations page lists none for GitHub offerings as of April 3, 2026. The filter remains optional.
• Is Windsurf still a separate product?No. Cognition renamed it Devin Desktop on June 2, 2026. It shares Devin’s pricing and terms.
• Which AI coding agent keeps prompts in the EU?Copilot supports EU in-region inference through GitHub Enterprise Cloud with data residency. Kiro processes enterprise inference inside its Europe geography. Cognition offers a dedicated VPC for the agent’s devbox.