AI Policy

Last updated: October 2026

In line with the transparency obligations of the EU AI Act (article 50), here is how NextWishlist uses AI.

Where we use AI

• Daily and weekly reports: an AI model writes, in plain language, findings that our code has already calculated (wishlist changes, post performance, deadlines), and picks 3 actions from rule-based recommendations.
• Classification: sorting your posts by type (gameplay clip, devlog, meme…) and analysing public Steam pages (style, title, AI disclosures) to build benchmarks.

All AI-generated content is labelled as such in the product and in emails.

What AI does not do

• It does not compute metrics, attribution or estimates: deterministic, testable code does.
• It never posts or acts on your accounts, and never makes decisions with legal or financial effect.
• It gives no legal, tax or financial advice.

Data sent to AI models

Only numerical summaries about your game and public information. Never your API keys, tokens, passwords, or another user's data. Numbers in generated text are automatically checked against the source data.

Training

We never train AI models on your data, and our AI provider does not train on data received through its business API under its commercial terms.

Limits

AI can be wrong or over-interpret a correlation. That is why every recommendation shows its basis: sample size, confidence level, source and date of the rule. You can flag any error with "Not relevant" or by writing to contact@nextwishlist.com.

AI-generated marketing assets and Steam

If you use AI to create marketing assets or Steam page content, Valve requires you to disclose it in Steam's content survey (rules updated on January 16, 2026). NextWishlist reminds you when relevant and will always encourage accurate disclosures.