kōdōkalabs
AI Ethics & Safety Statement

Innovation with Integrity:
Our Pledge to
Responsible AI

At kōdōkalabs, we believe that Generative AI is the most powerful engine for agency growth in history—but powerful engines require sophisticated brakes and steering.

As an agency serving enterprise clients, B2B SaaS leaders, and regulated industries, we adhere to a strict ethical framework regarding how we use, deploy, and manage Artificial Intelligence.

This statement outlines our commitment to Transparency, Data Sovereignty, and Human Accountability.

1. Data Sovereignty & Privacy
(The "No-Train" Policy)

We respect the confidentiality of your proprietary data.

The primary concern of our clients is intellectual property leakage. We strictly adhere to the following protocols:

Zero Training on Client Data: We configure our API connections (via OpenAI Enterprise, Anthropic, and Perplexity) to opt-out of model training. Your inputs, strategy documents, and customer data are never used to train public models.
Sandboxed Environments: When analyzing sensitive data (e.g., log files, GSC data, customer reviews), we use isolated, stateless environments where data is processed and immediately discarded after the session.
NDA Compliance: Our AI agents are bound by the same confidentiality agreements as our human staff. We do not input PII (Personally Identifiable Information) into LLMs unless a specific Data Processing Agreement (DPA) is in place.

2. Human-in-the-Loop (HITL)
Guarantee

AI generates. Humans verify. You publish.

We fundamentally reject the “auto-pilot” model of content generation. We believe fully automated content is a liability to your brand reputation and domain authority.

The 100% Review Rule: No asset—whether a blog post, a technical audit, or a strategy roadmap—leaves kōdōkalabs without being reviewed, fact-checked, and signed off by a human expert.
Fact-Checking Rigor: AI models can hallucinate. Our editors are trained to cross-reference every statistic, quote, and data point generated by AI against primary sources.
Nuance Injection: We use humans to inject brand voice, cultural nuance, and emotional intelligence that current AI models cannot replicate.

3. Transparency & Disclosure

We don’t hide the robot.

We are proud of our hybrid model and believe transparency builds trust.

Methodology Disclosure: We inform all clients that we utilize AI tooling for research, drafting, and data analysis. We do not present AI-generated drafts as purely human-written work during the draft phase.
Watermarking (Internal): We maintain an internal log of which tools and prompts were used to generate specific assets, ensuring we can audit the “chain of thought” if questions arise later.

4. Intellectual Property (IP)
Ownership

You own the output.

Work for Hire: Despite the use of AI tools, kōdōkalabs assigns full Intellectual Property rights of the final deliverables to the Client upon payment, exactly as we would with human-generated work.

Copyright Advisory: We advise clients on the current legal landscape regarding AI copyright (e.g., the inability to copyright purely raw AI output in some jurisdictions) and ensure our “Human-in-the-Loop” process adds sufficient human creativity to support copyright claims where possible.

5. Bias Mitigation & Fairness

We engineer against bias. Large Language Models inherit the biases of their training data. We actively prompt against and review for:

Stereotypes: Ensuring content does not reinforce harmful gender, racial, or cultural stereotypes.
Diversity of Thought: actively prompting our research agents to seek diverse sources and perspectives, rather than just the most dominant viewpoints in the training set.

kōdōkalabs AI Ethics & Safety Statement

Innovation with Integrity:Our Pledge toResponsible AI

1. Data Sovereignty & Privacy(The "No-Train" Policy)

2. Human-in-the-Loop (HITL)Guarantee