Product
AI Disclaimer
AI Output is a Starting Point
ADRITHM uses AI to generate ad copy, images, video scripts, voiceover, and recommendations. AI predicts likely text, images, or video — it doesn't know things or check facts. It can be wrong in ways that look right.
You must review everything before it goes live. What you publish is yours.
What AI Gets Wrong
- Factual errors: Invented statistics, wrong prices, non-existent features, made-up awards
- Compliance issues: Superlatives, health/financial claims not compliant with law or ASCI Code
- Similarity: Output may resemble existing work, taglines, or trademarks
- Bias: Training data contains bias. Output can reflect it in imagery, language, and depiction
- Visual errors: Generated images have known artifacts — hands, text, logos, product details
- Voice/video errors: Mispronounced names, wrong timing, incorrect on-screen text
- Inconsistency: Same prompt produces different results. Output is not reproducible.
What ADRITHM Does NOT Do
ADRITHM does not:
- Fact-check output
- Check against advertising law, the ASCI Code, or sector regulation
- Screen for copyright or trademark infringement
- Verify claims about your product, pricing, or performance
- Guarantee originality or that identical output won't be generated for someone else
- Guarantee any advertising outcome
The approval gate is not optional. Every AI-generated asset must pass human review and approval before going live. This is recorded.
AI Providers & Data Training
| Function | Provider | Your Data Excluded From Training? |
|---|---|---|
| Text generation | Anthropic, Google | ✓ Yes — confirmed |
| Image generation | Google (Imagen), Black Forest Labs (FLUX) | Google: ✓ Yes FLUX: ⚠ Pending |
| Video generation | Google (Veo) | ✓ Yes — confirmed |
| Voice generation | ElevenLabs | ⚠ Pending |
Important: Don't paste passwords, financial details, health data, or anyone else's confidential information into an AI prompt.