How YTForge uses AI — responsibly

1. Overview

YTForge ("we", "us") builds AI-powered tools that help YouTube creators research, write, design, and analyze their content. This AI Policy explains, in plain language, how our AI features work, which models power them, how your data is processed, and the rights you retain when you use them. It complements our Privacy Policy and Terms of Service; where anything is unclear, the Privacy Policy is authoritative on data handling.

Our north star is simple: AI should amplify creators, not replace them and not exploit them. Every feature is designed with a human in the loop — you review, edit, and approve before anything ships.

2. What We Build With AI

YTForge uses AI across four categories of tools:

• Generative writing: title ideas, full scripts, descriptions, hashtags, and tags — produced from your inputs (topic, niche, tone, length).
• Visual generation: AI thumbnails and short-form video concepts from text prompts and reference imagery.
• Analysis: SEO audits, channel analytics, retention analysis, and monetization eligibility checks that interpret public and provided data.
• Conversational assistance: the YTForge AI Chat that answers creator questions, suggests next steps, and drafts content.

Every AI output is labeled as AI-generated inside the product, and every output is editable before you use or publish it.

3. AI Models & Providers

We don't build foundation models from scratch. We route each task to the model best suited for it, using a mix of first-party fine-tunes and third-party providers:

• Large language models: used for scripts, titles, descriptions, chat, and SEO analysis. We select per-task based on quality, latency, and cost.
• Image models: used for AI thumbnail generation and visual concepts.
• Transcription models: used for the AI Transcript tool to convert audio to text and translate.
• First-party models: trained on a curated corpus of 4.2 million publicly accessible viral YouTube videos for trend detection and title/CTR scoring.

We evaluate every provider against security, data-handling, and output-quality criteria before integration, and we retain the right to switch providers as the field evolves. Provider APIs are bound by contracts that prohibit them from training on YTForge user data.

4. Your Data & AI Processing

When you use an AI tool, your inputs (prompts, channel URLs, uploaded images, script drafts) and the model's outputs are processed to deliver the result to you. The specifics:

• Real-time processing: inputs are sent to the relevant model provider, processed, and returned to you in your session.
• Storage: your generated outputs are stored in your account so you can access them later; you can delete any output at any time.
• Provider retention: our providers' default API terms do not retain your inputs beyond the processing window, and do not use them to train their models.
• Encryption: all data in transit uses TLS 1.3; data at rest uses AES-256.

For the full data-handling picture, see our Privacy Policy.

5. Training & Your Content

This is the part most creators care about, so we'll say it directly: we never train our public AI models on your private content.

Our first-party models are trained exclusively on a curated dataset of publicly accessible viral YouTube videos plus licensed third-party datasets. Your private scripts, prompts, channel data, and uploaded images are never included in training data — not in raw form, not in aggregated or anonymized form.

From time to time we invite Pro and Enterprise customers to opt in to research programs that may use their content to improve specific features. Participation is voluntary, fully transparent, revocable at any time, and off by default.

6. AI Output & Ownership

You own the AI output you generate on YTForge. We don't claim rights to your scripts, titles, thumbnails, or analyses beyond what's required to operate the Service for you.

• You can use it commercially: AI-generated output is yours to publish, monetize, and modify.
• You're responsible for review: AI can produce inaccurate, biased, or infringing content. Always review before publishing.
• Provenance: we encourage labeling AI-assisted content where appropriate, in line with platform and regional requirements.

Because AI output can reflect patterns in training data, you should verify facts, quotes, statistics, and any claim that could harm someone if wrong.

7. Limits & Human Oversight

AI is powerful but imperfect. To keep you in control:

• Human review gate: no AI output auto-publishes. You always review and approve.
• Fact-check prompt: for analytical tools, we surface a reminder to verify any specific statistic or claim.
• Usage limits: free and paid tiers have daily generation caps to prevent abuse and runaway costs.
• Quality feedback loop: the rating widget on each output trains our routing logic (not your content) to pick better models over time.

8. Your Rights

You have the right to:

• Access the AI outputs stored in your account.
• Delete any output or your entire account at any time.
• Opt out of optional research programs (they're off by default anyway).
• Withdraw any consent you've given for AI processing of your data.
• Lodge a complaint with your local data-protection authority.

Exercise most rights in one click from your dashboard, or email ai@ytforge.app.

9. Safety & Abuse Prevention

We don't allow YTForge's AI to be used to generate:

• Content that sexualizes minors or exploits anyone.
• Deceptive deepfakes or impersonation of real people without consent.
• Disinformation, election manipulation, or coordinated inauthentic behavior.
• Content that infringes others' copyright or trademarks.
• Content that promotes violence, terrorism, or self-harm.

We combine automated filters, provider-level safety models, and human review of flagged outputs. Violations can result in output blocking, account suspension, or termination. If you see harmful AI output from YTForge, report it to safety@ytforge.app.

10. Changes to This Policy

AI moves fast, and this policy will evolve with it. We'll notify you of material changes (new providers with different data terms, changes to training practices) via email and an in-product banner at least 30 days before they take effect. The "Last updated" date at the top reflects the most recent revision.

11. Contact

Questions about how YTForge uses AI?

• AI team: ai@ytforge.app
• Safety reports: safety@ytforge.app
• Privacy / DPO: privacy@ytforge.app