Version 2.1, effective 2026-10-23.
This Fair Use Policy is incorporated by reference into the PrivateMind Terms of Service. It contains operational and acceptable-use rules. Contractual rights, payment, liability and termination are governed by the Terms of Service. If this Policy conflicts with the Terms of Service, the Terms prevail; on data-protection matters the DPA and AI Privacy Policy prevail.
1. Introduction
This Fair use policy applies to all Client users, service accounts, demo and evaluation accounts, and integrations accessing PrivateMind, whether via the webchat interface or the API.
Fair use limits are necessary to protect the integrity of the service and ensure a fair and equitable level of service to users.
Clients are expected to refrain from behaviour that can reasonably be expected to cause service disruption.
2. Usage limits & Thresholds
2.1 Token Limits
- Demo Service: $10 of usage per user per month
- Enterprise: Custom limits available
2.2 Usage Enforcement
Clients accept that by using PrivateMind, for purposes of monitoring compliance with the Fair use policy Options Technology may monitor:
- Usage patterns and rate limit adherence
- Content quality and policy violations
- System load and resource allocation
- Suspicious or anomalous behaviour
- Content quality review is limited to automated screening: always-on for demo/evaluation accounts; for contracted Clients, content-level abuse monitoring is available on an opt-in basis as described in the AI Privacy Policy. Options does not manually review Client prompts or outputs except at the Client's request
2.3 Commercial Restrictions
Clients shall not resell PrivateMind, or offer its AI-generated content as a service.
Clients shall not use PrivateMind for bulk generation of content to train competing models.
3. Security Requirements
3.1 Credential Hygiene
Multi-factor authentication (MFA) is required for webchat access. Where the Client connects via its own identity provider (SSO/SAML), enforcement of MFA is the Client's responsibility; where Options Technology provides authentication directly, Options Technology enforces it.
API keys must be kept secure and must never be embedded in client-side code, mobile applications, or any publicly accessible repository.
Clients are responsible for rotating API keys on a regular basis and immediately upon any suspected compromise. API key expiry can be set in PrivateMind.
Any suspected credential compromise must be reported to Options Technology without undue delay via the incident reporting channel described in Section 3.4.
3.2 No Unauthorised Security Testing
Clients must not conduct penetration testing, vulnerability scanning, adversarial probing, jailbreak attempts, or prompt-injection attempts, or any attempts targeting other tenants, without Options Technology's prior written authorisation.
Attempts to extract model weights, training data, or system prompts, or to otherwise reverse-engineer the underlying model or infrastructure, are prohibited.
Good-faith security researchers may report vulnerabilities through Options Technology's responsible disclosure channel; Options Technology will not treat good-faith, coordinated disclosure as a breach of this clause.
3.4 Incident Reporting
Options Technology to Client: Options Technology will notify affected Clients of a confirmed security incident affecting the service without undue delay, and in any event within 48 hours of confirmation, consistent with applicable breach-notification norms.
Client to Options Technology: Clients must notify Options Technology without undue delay of any security incident on their side that may affect the service; including but not limited to leaked credentials, unauthorised access to their environment, or compromised integrations.
The incident reporting channel and contact details are: privatemindsupport@options-it.com.
4. Data Inputs
Unless agreed in writing as part of an Enterprise License, Clients are advised not to submit the following to PrivateMind via webchat or API:
- Special category personal data (e.g., health, biometric, or similarly sensitive data under UK/EU GDPR Article 9)
- Authentication credentials, secrets, private keys, or tokens
- Payment card or other regulated financial account data
- Source code, trade secrets, or confidential information belonging to a third party, unless the Client has the right to disclose it
Clients remain solely responsible for the content they submit and for ensuring it is permitted under their internal & external policy requirements.
5. Compliance & Data Protection
5.1 Controller / Processor Allocation
The allocation of data controller and data processor responsibilities between Options Technology and the Client is set out in the Privacy Policy.
Clients remain responsible for ensuring they have a lawful basis for any personal data submitted to PrivateMind.
5.2 Encryption
Data is encrypted in transit via secure protocols and at rest (AES-256-GCM).
5.3 No Training on Client Data
Options' commitments on the use of Protected AI Data, including that it is never used to train, fine-tune or improve any model except with the Client's prior written consent, are set out in the AI Privacy Policy.
5.4 Sub-Processor and Infrastructure Transparency
Options Technology will make available, on request, a current list of sub-processors and underlying model/infrastructure providers used to deliver PrivateMind, including relevant data residency information.
Where personal data is transferred outside the UK/EEA, Options Technology relies on appropriate safeguards (e.g., UK GDPR Standard Contractual Clauses / International Data Transfer Agreement) as set out in the DPA.
5.5 Regulated-Industry Responsibility
Clients remain solely responsible for their compliance with regulatory obligations in regards to any inputs submitted into PrivateMind or the reliance on any data produced by PrivateMind, including but not limited to those under FCA, SEC, MAS.
Clients should assess whether content generated via PrivateMind may constitute a business record for the purposes of their record-keeping obligations (e.g., SEC Rule 17a-4, MiFID II, FCA SYSC) and retain it accordingly.
Clients should independently confirm that reliance on AI-generated output is compatible with their industry and internal policy requirements before acting on it, particularly where regulatory guidance on generative AI outputs is still evolving.
5.6 EU AI Act Transparency
Consistent with Article 50(1) of the EU AI Act, users of PrivateMind's webchat are informed that they are interacting with an AI system. This disclosure is built into the product interface and reinforced here.
AI-generated content is also marked as artificially generated in a machine-readable way, consistent with Article 50(2) of the EU AI Act.
5.7 Sanctions and Export Control
Access to PrivateMind is not permitted from prohibited or sanctioned jurisdictions, or by sanctioned parties, in each case as determined under applicable export control and sanctions law.
5.8 Third Party Policy Compliance
Client agrees to comply with the licence terms and usage policies of the third-party models made available within PrivateMind. A current list of served models, their publishers and applicable licence terms is available within the PrivateMind interface. Models may be added, substituted or removed on at least 7 days' notice. Where a model's licence requires attribution, that attribution will appear in the interface.
6. Ethics & Responsible Use of AI
6.1 Prohibited Use Cases
PrivateMind must not be used to:
- Engage in, or facilitate, unlawful activity of any kind
- Impersonate other individuals or entities
- Generate malware, phishing content, or content designed to compromise the security of any system
- Create or spread disinformation, or engage in market manipulation or insider dealing
- Produce content facilitating self-harm, child sexual exploitation, or weapons (including chemical, biological, radiological, or nuclear) development
- Make, or materially inform, decisions with legal or similarly significant effect on individuals, such as credit, employment, or insurance decisions, without qualified human review
6.2 Mandatory Human Oversight
Outputs from PrivateMind are assistive and advisory in nature, not authoritative. The platform must not be used for fully automated decision-making without human review.
Outputs are not a substitute for professional legal, financial, investment, medical, or other regulated advice.
6.3 Known Limitations Disclosure
PrivateMind, like all AI systems, may produce inaccurate, incomplete, or fabricated ("hallucinated") output, may reflect bias, and has a knowledge cut-off date beyond which it lacks awareness of events.
Clients should independently verify outputs before relying on them for consequential decisions.
6.4 No Misrepresentation
Where legally or contextually required, Clients must disclose that content was generated or assisted by AI.
6.5 Reporting Harmful or Concerning Outputs
Clients may report harmful, biased, or otherwise concerning outputs through Options Technology's feedback channel within PrivateMind itself or by emailing privatemindsupport@options-it.com.
Any concerns about a PrivateMind AI system or its behaviour can be reported at ai-concerns@privatemind.com.
Reports are reviewed as part of Options Technology's ongoing AI governance and continuous improvement process.
6.6 Output Ownership and Intellectual Property
As between Options and the Client, the Client owns the outputs generated through its authorised use of PrivateMind, to the extent permitted by applicable law. Options assigns any rights it may have in such outputs. Clients are responsible for how they use outputs, including clearing third-party rights where required.
7. Policy Updates
This policy is reviewed quarterly.
Major changes are announced 30 days in advance.
Changes that are administrative, clarifying or reflect operational improvements may take effect on posting; any change that materially reduces Client rights or adds new obligations takes effect on 30 days' notice.
Revision history
| Version | Effective | Changes |
|---|---|---|
| 1.0 | 2026-07-27 | Initial policy. |
| 2.1 | 2026-10-23 | Revised policy. Editorial correction 2026-10-29: the sub-processor list is available on request (5.4). |