AI Usage & Responsible AI Policy

1. Why this policy exists

Automation and AI overlap, and clients reasonably want to know where the line is: what we use, what happens to their data, and what we won't do. This page answers that. It applies to our own working practices and to systems we build for clients.

2. Where AI shows up in our work

AI may be involved in two distinct ways, and they're worth separating:

As a tool we use to do the work

We may use AI assistants to help draft code and configuration, review our own work, produce documentation, or think through an approach. We are responsible for everything we deliver regardless of how it was produced — output is reviewed, tested, and corrected by a person before it reaches you.

As a component inside what we build

Some automations use AI directly — classifying incoming email, extracting fields from documents, summarising, drafting replies. When a build includes an AI component, we tell you before we build it: which provider, what data flows to it, what it costs, and what happens when it gets something wrong.

3. Your data and model training

• We do not use your data to train AI models, and we don't permit our providers to either.
• Where we use third-party AI services, we use business or API tiers configured so that submitted data is not retained for training.
• We don't paste client data into consumer AI tools or personal accounts.
• We minimise what's sent: an AI component receives the fields it needs, not whole records, wherever the design allows.
• AI providers used in your systems are treated as sub-processors under our Data Processing Agreement, and we'll name them.

4. Human oversight

We do not build fully autonomous systems that make consequential decisions about people. Where an automation touches hiring, firing, promotion, pay, credit, insurance, housing, education, immigration status, healthcare, legal outcomes, or access to essential services, we design it so a person makes the decision. AI may sort, summarise, or suggest; a human decides, and that decision is recorded.

For lower-stakes automation, we still design in review points, logging, and a way to see what the system did and why.

5. What AI is and isn't good at

Language models are good at handling messy input, summarising, classifying, and drafting. They are unreliable at arithmetic, at anything requiring guaranteed precision, and at knowing when they're wrong. They produce confident, plausible output that is sometimes simply incorrect.

We'll tell you when a rules-based approach is the better answer, even when AI is the more impressive demo. Plenty of problems are better solved with a lookup table. Where we do use AI, we build for the error case: validation, confidence thresholds, fallbacks, and a route to a human.

6. Transparency

You will always know which parts of your workflow use AI — it's documented in the handover, not buried. Where an AI system communicates with your customers or employees, we recommend and will help you disclose that they're interacting with an automated system. We won't design systems intended to make people believe they're talking to a human when they aren't.

7. What we won't build

We decline work, including AI work, that is designed to:

• Impersonate a real person or organisation, or generate deceptive content presented as genuine — fake reviews, fabricated records, synthetic media of real people
• Produce disinformation, or content designed to manipulate elections or public health decisions
• Conduct surveillance of individuals without their knowledge, including covert monitoring of employees beyond what the law allows
• Score, rank, or profile people on protected characteristics, or serve as a proxy for doing so
• Send unsolicited bulk messaging, or evade platform rate limits, terms of service, or anti-abuse measures
• Scrape personal data in violation of a platform's terms or applicable law
• Generate sexual content involving minors, or content that harasses or endangers a specific person

This list isn't exhaustive. We reserve the right to decline any engagement we consider harmful, and to stop work if a project turns in that direction. See also our Acceptable Use Policy.

8. Your responsibilities

You remain responsible for how you use what we build, for the accuracy of the data you feed it, and for compliance obligations in your industry — including any rules about automated decision-making, disclosure, record-keeping, or professional oversight that apply to you. Where AI output is used in a regulated context, you're responsible for the review your regulator expects. We'll tell you what we know; we can't be your compliance function, and nothing here is legal advice.

9. Changes

This area moves quickly and we expect to revise this policy. The "Last updated" date above reflects the current version.

Contact

Questions about this policy:

LogicLoop LLC
1805 Robin Ave
Fort Worth, TX 76164
contact@logicloop.agency