Safe harbor for AI-curious builders.

1. What this community is for

Dabblewith.ai exists to publish practical, forkable AI workflows and to host honest conversations between builders, creators, researchers, healthcare/business ops teams, and no-code automators. It is not a place to push tools, evangelize doctrines, or grandstand. The default mode is "I tried this; here's what worked and what didn't."

2. Constructive AI debate

We welcome strong disagreement about models, tools, providers, prompting techniques, automation tradeoffs, and where AI should and should not be used. We expect that disagreement to be specific, evidenced, and aimed at the work — not at the person. Ideological ambushes, dunking, and "I told you so" energy don't belong here.

• Bring an example, not just a label.
• Disagree with the workflow, not the workflow's author.
• "This is naïve" is not feedback. "I'd change X because Y" is.

3. Respect for creators, artists, writers, and researchers

The people whose work AI was trained on are not abstractions to us. We will not publish workflows that scrape, impersonate, or undermine creators, artists, writers, or researchers, or that use AI to systematically erase attribution.

• Voice and style emulation must be grounded in your own work or work you have explicit license to use.
• Cite sources. AI-generated summaries must include verifiable claims, not paraphrases that erase the original.
• Don't ship a workflow that quietly removes a human's byline.

4. Privacy and data ownership

We treat privacy as a default, not a setting.

• Workflows on this site must say what data they handle and how that data is sanitized.
• Never paste a real customer or community member's name, phone number, email, message content, health details, or financial details into a third-party LLM without consent and a clear lawful basis.
• If a workflow handles regulated data (PHI, financial, legal, education records, minors), it must say so plainly and follow the relevant jurisdictional rules.
• This site's analytics never collect PII, phone numbers, raw message content, or contact details. See the privacy page.

5. No harassment, no doxxing

No targeted harassment, slurs, sexual content directed at individuals, threats, doxxing, or coordinated brigading. Disagree publicly with arguments, never with someone's identity.

6. Human oversight for AI moderation

If Dabblewith.ai uses AI to triage flags, route questions, or summarize discussion, every consequential action — removing a post, removing a member, escalating a report — is taken by a human, with the AI's reasoning visible to the human reviewer. We will publish when, where, and how AI assists in moderation, and members can request a human-only review of any moderation decision affecting them.

7. Bias, provenance, and data security

We expect every workflow to consider:

• Bias. Who could this workflow systematically disadvantage if it scales? What's the human-review point that catches it?
• Provenance. Where do the inputs come from? Is the source consented and licensed for this use?
• Data security. What happens to inputs after the workflow runs? Where do logs live? Who can read them? When are they deleted?

8. What we will not automate

Some things stay with humans on this platform, period:

• Clinical and medical decisions.
• Legal advice that creates duties or liabilities for a real person.
• Financial advice with real-money consequences.
• Hiring/firing decisions, performance reviews, or actions affecting protected categories.
• Moderation actions against members, including removal, restriction, or escalation.
• Anything that would be illegal in the user's jurisdiction.

We may help you redesign a workflow so that AI assists humans in these areas — but we will not publish, host, or promote AI workflows that take these actions autonomously.

9. Self-promo & commerce

Builders are welcome to share their tools, products, and services in context — when they help solve a specific problem someone raised. We do not allow drive-by promo, recruiter cold pitches, or affiliate stacks disguised as workflows. Sponsored content must be disclosed.

10. Reporting and appeals

If you see something off — a workflow that violates this policy, a comment that crosses a line, or a moderation decision that doesn't sit right — email bhuvanesh@neuralhawk.ai with the subject "Community policy report". A human reads every report. Appeals get a human-only review.