Artificial Intelligence Policy
Effective September 5, 2026. Artificial intelligence is a normal part of how Redstart Media researches, designs, writes, builds, tests, connects, and improves modern digital work. This policy explains where AI helps, where people remain responsible, and how we separate protected information from content intentionally published for discovery.
1. Purpose and scope
This policy applies to AI-assisted work performed by Redstart Media, Inc., including generative AI, machine learning, AI-assisted analysis, automation, software development, content production, media workflows, search and discovery, and integrations with third-party systems. A client agreement may impose stricter project-specific requirements, and applicable law always controls.
2. How we use AI
We use AI to explore ideas, organize research, draft and refine content, generate or edit creative materials, analyze information, assist software development, test applications, automate repeatable steps, connect systems, improve accessibility, and help teams move from concept to working application more quickly. AI can reduce repetitive labor and shorten feedback cycles, but speed is never the only measure of a successful result.
3. Human ownership
No fully autonomous consequential systems. Redstart Media does not deploy systems that independently make consequential business, employment, financial, legal, safety, or customer decisions without defined human responsibility. Every AI-enabled workflow we operate has a human owner. The depth of review is proportional to the risk, context, audience, reversibility, and impact of the work.
4. Human review
People set the objective, choose the tools and information, review material outputs, resolve uncertainty, and remain accountable for what is delivered. We may require subject-matter review, source checking, testing, approval, or a manual release gate before an AI-assisted result reaches a public or production environment. Human involvement does not make an output perfect, so we also design for correction, escalation, and improvement.
5. Quality and lifecycle value
AI is most useful when it improves the complete life of a solution. We use human review, reusable systems, documentation, testing, maintainable architecture, and measured automation to reduce rework and operating friction. This approach is intended to improve quality while lowering avoidable total cost of ownership—not merely to produce a first draft faster.
6. Accuracy and limitations
AI systems can produce incomplete, incorrect, outdated, biased, or fabricated material. We do not treat an AI output as verified simply because it sounds confident. Review may include comparison with authoritative sources, direct inspection, code validation, factual checks, accessibility review, rights review, and testing appropriate to the use. We communicate material limitations when they affect how a client or user should rely on a result.
7. Fairness, accessibility, and harm
We consider whether an AI-assisted system or artifact could create unfair exclusion, reinforce harmful bias, mislead an audience, impersonate a person, reduce accessibility, or create a disproportionate risk for a person or group. Higher-impact uses receive stronger safeguards. We do not knowingly use AI to facilitate unlawful discrimination, deception, harassment, exploitation, or manipulation.
8. Customer and user data
We do not train AI models on customer or user data. We do not sell customer or user data to AI companies or make private client information available for another party's model training. Information supplied through an inquiry, account, project, private system, or authenticated experience remains governed by our Privacy Policy, the relevant client agreement, and the purpose for which it was provided.
9. Data minimization and approved tools
When AI assistance is appropriate, we aim to use only the information needed for the task and to avoid unnecessary personal, confidential, regulated, credential, or security-sensitive data. We favor service configurations and commercial terms that restrict provider training and secondary use. Third-party AI providers may process submitted information to deliver their service, so tool selection, access, retention, and contractual protections are evaluated according to the sensitivity of the work.
10. Public content and AI discovery
Public website content is different from private customer data. By default, Redstart does not block legitimate search and AI crawlers from material deliberately published to the open web. We may configure Redstart sites—and client public-facing sites when authorized—to make that public content eligible for indexing, retrieval, summarization, citation, and, where a provider's rules and applicable law permit, model training. This can improve visibility in AI-assisted research, recommendation, discovery, and future marketing channels, but it does not guarantee placement, targeting, leads, or revenue.
11. Client control of public access
Public-content discoverability is implemented subject to the client's instructions, contracts, publishing rights, and the technical controls available on the site and platform. A client may request a different crawler policy. We can use recognized controls such as robots directives and metadata where appropriate, while acknowledging that technical directives do not create a universal legal license, cannot bind every automated actor, and cannot guarantee that previously collected public material will be removed from external systems.
12. Intellectual property and creative work
We consider ownership, license terms, attribution, likeness, confidentiality, and source provenance when selecting tools and delivering AI-assisted work. AI involvement does not erase the rights of creators, clients, or third parties. We do not knowingly present another person's protected work as original Redstart or client work, and we use human creative direction to shape, select, revise, and approve deliverables.
13. Security, retention, and access
AI-enabled workflows remain subject to reasonable security, access-control, and retention practices. We limit access according to role and need, avoid placing secrets or credentials into general-purpose AI tools, and retain working material only as long as reasonably needed for delivery, support, records, security, or legal obligations. No system is risk-free, so safeguards are reviewed as tools and threats change.
14. Transparency and questions
We are open about the fact that AI is used throughout our practice. We provide more specific disclosure when AI materially affects a person's interaction, a consequential recommendation, a synthetic representation, or a deliverable where that context would reasonably matter. Clients may ask how AI was used in their work, identify sensitive boundaries, request additional review, or raise a concern through Contact.
15. Governance and standards
Our approach is informed by the NIST AI Risk Management Framework, the OECD AI Principles, and current privacy and consumer-protection guidance. These references guide risk-based governance, transparency, accountability, security, privacy, and human-centered oversight; they do not represent a certification or warranty.
16. Changes and contact
AI capabilities, provider terms, and laws change quickly. We may revise this policy as our services, tools, risks, or obligations change, and the effective date above identifies the current version. Questions or concerns may be sent through our contact page or to privacy@redstartmedia.com.