Joint ventures and consortium projects create shared systems that no single partner fully owns, and that is exactly where access, data and accountability go unmanaged. Partners connect networks, share a cloud tenant or project workspace, exchange data under a mix of contracts and second staff who keep accounts in both organisations. When the venture ends, or one partner leaves, the shared environment often keeps running with credentials, copies of data and trust relationships nobody remembers granting. This guide covers the security of shared environments in joint ventures and multi-party consortia: governance and decision rights, the choice of hosting model, identity and access across partners, data ownership and handling, joint monitoring and incident response, and a controlled exit. It is for security leads, IT and identity architects, programme managers and legal or commercial teams setting up or running a venture. Sharing intelligence between competitors for fraud prevention is a different problem, covered in the Fraud Data Sharing and Consortium Intelligence Guide.
Source note: NIST SP 800-47 Rev. 1 and ICO points below are summarised from the official texts listed in the references, as checked in November 2026. This is security guidance, not legal advice. Confirm against the official text before relying on it.
Key takeaways
- Agree security decision rights before the first connection. Name who approves access, accepts risk, leads incidents and authorises disconnection for the shared environment.
- Choose the hosting model deliberately. A neutral environment owned by the venture is easier to govern and exit than one hosted inside a partner's estate.
- Every account and credential needs a home organisation and an end date. Federate partner users from their own identity providers and give every service account and API key an owner on one side.
- Data ownership must be written down per data set. Contributed, jointly created and personal data each need different rules for use, retention and return or destruction.
- Plan the exit at the start. Termination, partner withdrawal and emergency disconnection should be procedures with steps and owners, not a clause that says "the parties will cooperate".
Why shared environments are different
Ordinary third-party risk assumes one customer and one supplier, with the customer setting the controls. In a joint venture or consortium there are two or more principals with equal standing, a shared environment that may belong to a separate legal entity, and staff whose loyalties and accounts span organisations. Several problems follow:
- No default owner: each partner assumes another is patching, reviewing access or watching logs.
- Uneven maturity: the environment is only as strong as the weakest partner's identity provider, endpoints and supplier chain, because each partner's users and integrations connect to it.
- Competitive sensitivity: partners may compete outside the venture, so information barriers matter alongside external threats.
- Changing membership: consortia add and lose members over a project's life; each change is a joiner or leaver event at organisational scale.
NIST frames any exchange of information between organisations as a lifecycle in SP 800-47 Rev. 1, Managing the Security of Information Exchanges: plan, establish, maintain and discontinue. That structure fits a joint venture well and is used below.
Governance and decision rights
Create a joint security function or at least a joint security working group reporting to the venture's steering committee, with a named security lead for the shared environment. Record in a responsibility matrix who does what. SP 800-47 Rev. 1 recommends a joint planning team, protection requirements agreed up front and approval of the exchange by each organisation's authorising official.
| Decision or activity | Typical owner | Must consult | Record in |
|---|---|---|---|
| Security policy for the shared environment | Venture security lead | Each partner's CISO | Venture security policy, approved by steering committee |
| Granting partner user access | Data or system owner in the venture | Sponsoring partner manager | Access request system of the hosting party |
| Risk acceptance for exceptions | Steering committee or delegated venture executive | Partners whose data is affected | Venture risk register |
| Incident lead for the shared environment | Hosting party's incident team | All partner incident contacts | Joint incident response plan |
| Emergency disconnection of a partner | Venture security lead or predesignated deputy | Steering committee, legal | Disconnection procedure with written criteria |
| Regulatory notifications | Each controller or regulated entity for its own obligations | Other partners, counsel | Notification matrix |
Write the security commitments into the joint venture agreement or a security schedule attached to it: minimum controls each partner maintains for users and systems that connect, audit and assessment rights, incident notification windows between partners, and the exit procedure. A template for the data side is the Data Sharing Agreement Security Schedule Template.
Choosing a hosting model
| Model | Description | Strengths | Exit implications |
|---|---|---|---|
| Hosted by a lead partner | Shared workspace, applications and data live in one partner's tenant; others are guests | Fast to set up; uses lead partner's mature controls | Other partners depend on the host for data return; host must remove guests and integrations |
| Neutral venture environment | Separate tenant, cloud accounts and identity provider owned by the venture entity | Clear ownership; uniform policy; cleanest exit or transfer | Environment can be handed to the surviving entity or wound down as a unit |
| Federated, each partner hosts its part | Each partner keeps its systems; interconnections and APIs link them | Data stays at home; minimal shared estate | Exit means closing interconnections and revoking cross-organisation credentials |
| Third-party collaboration platform | A provider's data room or project platform used by all partners | Provider manages platform security; built-in audit | Depends on the provider's export and deletion features and contract |
Decision guidance (NHI Mgmt Group recommendation): short projects with limited sensitive data suit a lead-partner host or a third-party platform. Long-lived ventures that may be spun out, sold or bought by one partner suit a neutral environment from day one, because migrating later is costly. Consortia with many members and strict data sovereignty needs suit the federated model. Large partners with mature tooling should resist the temptation to host everything simply because they can: hosting makes them responsible for every other partner's access.
Identity and access across partners
- Federate rather than duplicate: partner users authenticate with their home identity provider through SAML or OIDC federation or cross-tenant guest access, so their accounts are disabled when they leave their employer. Require MFA, and decide whether to trust partner MFA and device compliance claims or enforce your own. See the Federation and Trust Relationships Guide and, for Microsoft environments, the Entra ID Cross-Tenant Access and B2B Guest Governance Guide.
- Sponsor every external account: each partner user has a named sponsor in their own organisation who confirms the need at each review. In NHI Mgmt Group's view, expiring access by default (for example 90 or 180 days, renewable) prevents silent accumulation.
- Handle secondees explicitly: staff seconded into the venture often get a venture account and keep their partner account. Record both, decide which one has access to which data, and make the secondment end date an access end date.
- Information barriers: where partners compete, group-based access by partner organisation and data set prevents one partner's staff seeing another's contributed data. Apply it to search, shared drives and reporting, not just applications.
- Privileged access: limit administration of the shared environment to a small named group, use dedicated admin accounts, and never share an admin credential between partners. See the Cross-Account and Cross-Tenant Access Guide.
- Reviews: run access reviews at least quarterly with each partner confirming its own users. See the How to audit guest and external users guide.
Where identity fits: non-human identities in the venture
The credentials most likely to survive the venture are not people's. Integrations between partner systems and the shared environment run on service accounts, API keys, OAuth applications, workload federation trusts and data pipeline credentials. A partner's engineer creates a key, embeds it in a job and moves on; years later the key still pulls data out of the shared environment.
- Register every non-human identity in the venture with its owning partner, purpose, scopes and expiry.
- Prefer short-lived, federated workload credentials over long-lived keys for cross-organisation integrations.
- Scope each integration to the minimum data sets, read-only where possible.
- Include non-human identities in quarterly reviews and in the exit inventory.
See the NHI Offboarding and Decommissioning Guide and the API Key Management Guide.
Data ownership and handling
Disputes at exit are usually about data, so classify it at the start and record ownership per data set.
| Data category | Ownership | Use rules | At exit |
|---|---|---|---|
| Contributed data | Contributing partner | Use only for the venture purpose; no onward sharing | Returned or destroyed with certificate |
| Jointly created data and results | As set out in the venture agreement | Per agreement; often licensed to all partners | Copies distributed or retained as agreed |
| Personal data | Controller or joint controllers as determined | Purpose limitation, transparency, rights handling agreed | Deleted, returned or transferred lawfully |
| Security logs and telemetry | Hosting party or venture entity | Available to all partners for incident response | Retained for agreed period to support investigations |
| Encryption keys | Party that owns the protected data | Partner-held keys where data must stay under partner control | Revoked or destroyed to render remaining copies unreadable |
- Joint controllers: where partners jointly decide the purposes and means of processing personal data, they are joint controllers. The ICO's guidance says they need a transparent arrangement setting out who carries out each controller obligation, should make its main points available to individuals, and that each joint controller remains liable for the entire damage unless it proves it is not responsible.
- Labelling: tag data with its owning partner and classification at creation so access rules and exit sorting can be automated. See the Data Classification Guide.
- Keys: where a partner contributes highly sensitive data, encrypting it with keys that partner controls gives it a clean revocation lever at exit. See the BYOK, HYOK and External Key Management Guide.
Joint monitoring and incident response
- Logging: agree which logs the shared environment produces (authentication, admin actions, data access and export, integration activity), where they are held and how each partner can query them. SP 800-47 Rev. 1 lists incident reporting and event logging among the system security considerations in an interconnection security agreement.
- Detection: the hosting party monitors the shared environment; each partner monitors its own users and endpoints that connect. Agree how a partner tells the venture that one of its users or systems is compromised.
- Notification between partners: set a window (for example 24 hours from confirmation) for telling other partners about incidents that may affect shared systems or contributed data, shorter than any partner's regulatory clock.
- Joint exercises: run a tabletop at least yearly covering compromise of a partner identity provider and an insider exporting contributed data.
Exit, withdrawal and emergency disconnection
SP 800-47 Rev. 1 distinguishes planned discontinuance, with written notice, a timeline, coordinated disposition of shared data and updated security documentation afterwards, from emergency discontinuance in response to an attack, where the system owner, or a predesignated staff member under written criteria if the owner is unavailable, can authorise immediate disconnection, notifies the other party's emergency contact and preserves evidence. It also advises consulting legal counsel well before an emergency disconnection might be needed. NHI Mgmt Group recommends turning these into three procedures; the steps and example periods below are our recommendations.
Planned exit or venture wind-down
- Freeze: agree a cut-off date; stop new access grants and new integrations.
- Inventory: list every human account, guest, secondee account, non-human identity, federation trust, interconnection, data store and key in the shared environment, with owning partner.
- Sort and return data per the ownership table, using labels; record what was returned, copied or retained and by whom.
- Revoke access: disable partner accounts and guests, revoke API keys and OAuth grants, remove federation and cross-tenant trusts, close network interconnections and VPNs.
- Destroy remaining copies using media sanitisation appropriate to the data, or destroy partner-held keys, and issue destruction certificates.
- Verify: check sign-in and integration logs for any activity after the cut-off date for at least 30 days, then retain logs for the agreed period.
Partner withdrawal
Run the same steps for the departing partner's accounts, integrations and data only. Re-key any shared secrets the departing partner knew, and review whether jointly created data the partner retains needs continuing confidentiality obligations.
Emergency disconnection
Pre-agree the criteria (for example confirmed compromise of a partner identity provider or active data exfiltration), who decides, the technical steps (block federation, disable partner groups, cut interconnections) and how the affected partner is told. Rehearse it.
Framework mapping
| Framework | Reference | Joint venture relevance |
|---|---|---|
| NIST SP 800-53 Rev 5 | CA-3 Information Exchange; AC-21 Information Sharing | Agreements for interconnections and shared data |
| NIST SP 800-53 Rev 5 | AC-20 Use of External Systems; PS-7 External Personnel Security; AC-2 Account Management | Partner systems, partner staff and secondee accounts |
| NIST SP 800-53 Rev 5 | MP-6 Media Sanitization | Destruction of shared data at exit |
| NIST CSF 2.0 | GV.RR-02; GV.SC-07; GV.SC-10 | Roles across partners; ongoing partner risk; activities after the partnership ends |
| NIST CSF 2.0 | PR.AA-01; ID.AM-07 | Identities managed for users and services; data inventories per data set |
| ISO/IEC 27001:2022 Annex A | 5.14 Information transfer; 5.19 to 5.22 supplier relationships; 8.10 Information deletion | Transfer agreements, partner obligations, deletion at exit |
Detection and measurement
- External accounts without a sponsor or expiry date: target zero.
- Non-human identities in the shared environment with a recorded owner and expiry: target 100%.
- Quarterly access review completion by each partner: target 100%, with revocations tracked.
- Data sets with recorded owner and classification: target 100% before exit planning begins.
- Days between a partner leaver event and access removal in the shared environment: should match the partner's own leaver SLA.
- Sign-ins or API calls from former partners after cut-off: target zero; any occurrence is an incident.
Common failure patterns
- Security left out of the term sheet: the venture agreement covers intellectual property and profit share but not access, logging, incident notification or exit, so there is nothing to enforce.
- Local accounts for partner staff: accounts created in the host directory instead of federation survive the person leaving their employer.
- Orphaned integration keys: API keys and service accounts created for a data feed keep working long after the project ends.
- Everyone sees everything: a single project group gives every partner access to every partner's contributed data, breaching confidentiality between competitors.
- Exit by contract only: the agreement says data will be returned or destroyed, but nobody inventoried where it lives, so copies remain in backups, exports and personal drives.
Practitioner checklist
- Name a venture security lead and agree a responsibility matrix covering access, risk acceptance, incidents and disconnection.
- Attach a security schedule to the venture agreement with minimum controls, audit rights, notification windows and exit steps.
- Choose the hosting model with exit in mind; prefer a neutral environment for long-lived ventures.
- Federate partner users, require MFA, assign sponsors and set expiry on all external access.
- Register every non-human identity with owner, scope and expiry; prefer federated workload credentials.
- Record ownership and classification per data set and label data by owning partner.
- Agree logging, joint monitoring and partner-to-partner incident notification; exercise yearly.
- Write and rehearse planned exit, partner withdrawal and emergency disconnection procedures.
- Monitor for any activity by former partners after cut-off.
Standards and references
- NIST: SP 800-47 Rev. 1, Managing the Security of Information Exchanges
- NIST: SP 800-47 Rev. 1 full text (lifecycle and agreement templates)
- NIST: The NIST Cybersecurity Framework (CSF) 2.0
- NIST: SP 800-53 Rev. 5, Security and Privacy Controls
- ICO: What does it mean if you are joint controllers?
- Microsoft Learn: B2B collaboration overview
Related NHI Mgmt Group resources: Fraud Data Sharing and Consortium Intelligence Guide · Federation and Trust Relationships Guide · NHI Offboarding and Decommissioning Guide · Third-Party Risk Management Guide