AI Usage Policy

Last updated: May 2026

This page sets out Game Hub Arena's position on the use of generative AI in catalog operations and on the use of Game Hub Arena content for AI training. Headline: limited operational AI use, no AI-generated game descriptions or ratings published as human-curated, no AI-generated game content presented as developer-authored, and explicit AI-training opt-out.

The five positions

Limited operational AI use is permitted for catalog research, foreign-language source-material translation as a working tool, and code-generation for non-content infrastructure.
AI-generated game descriptions are not used — game descriptions in the catalog are derived from developer / distribution-network metadata with editorial paraphrasing, not generated wholesale by AI.
AI-generated game ratings are not used — content ratings are applied by the operator with reference to the framework on Content Ratings.
AI-generated game content presented as developer-authored is prohibited. The Site does not host AI-generated games disguised as developer-made; the catalog is human-developer-made HTML5 games.
Game Hub Arena content is opted out of AI training. The opt-out is asserted contractually and implemented technically.

1. Where AI is used in operational support

Permitted operational uses:

Catalog research assistance. When researching a specific game's developer attribution, genre context, or comparable-games framing, AI can serve as a research-acceleration tool. The AI surfaces possible sources; the operator reviews and decides.
Foreign-language translation as a working tool. Some indie HTML5 games come with developer-provided descriptions in non-English languages; AI translation as a working aid for the operator with verification of accuracy where consequential.
Code-generation for non-content infrastructure. Site backend (CloudArcade customizations, automation scripts for sitemap generation, data-import helpers, dead-link checkers). All output reviewed before deployment.
Spell-check and grammar tools with integrated AI features for catalog-entry text.

2. Where AI is not used

The following uses are prohibited:

AI-generated game descriptions presented as the catalog's editorial description. Game descriptions are derived from developer-provided metadata or distribution-network metadata with editorial paraphrasing, not AI-generated wholesale.
AI-generated content ratings. Content ratings are operator decisions with reference to the rating framework.
AI-generated genre tagging at scale. Genre tags are human-curated against the catalog's controlled vocabulary.
AI-generated game content (the games themselves). The catalog is HTML5 games made by human developers. Games entirely AI-generated and presented as developer-authored are not legitimate catalog candidates.
AI-generated reviews or commentary content that would appear in the catalog as if human-written.
AI-templated reader-correspondence responses. Reader email is answered by the operator.
AI-generated developer attributions or credits. Developer credits come from the actual game's metadata; AI tools that confabulate developer attributions are not used.

3. Where AI workflows could fail in this context

Generative AI has well-documented failure modes that are particularly relevant to game catalogs:

Game-developer confabulation. AI tools can attribute games to wrong developers, particularly for indie games where the AI's training data is thin. Operator verification against actual game metadata is the mitigation.
Genre misclassification. AI tools may not recognize sub-genre boundaries; "puzzle-platformer" can be misclassified as "puzzle" or as "platformer" individually.
Content-rating inflation or deflation. AI may rate games on stylistic appearance rather than actual content; mature-themed pixel-art games may be rated too lenient, while cartoon-styled action games may be rated too strict.
Description-padding with unsourced claims. AI tools tend to add plausible-sounding but unverified detail; descriptions are paraphrased from actual developer-provided material, not expanded by AI.

4. AI training opt-out — expressly asserted

Use of any content on Game Hub Arena for the training, fine-tuning, evaluation, or other development of artificial-intelligence systems, machine-learning models, or large language models is expressly prohibited. This opt-out is asserted under:

EU Directive 2019/790, Article 4(3) (text-and-data-mining exception with reservation of rights), as transposed into Italian law by D.Lgs. 177/2021.
U.S. Copyright Act, 17 U.S.C. § 106.
Italian copyright law L. 633/1941.
Equivalent rights of authorship in other jurisdictions.

The opt-out applies to:

Crawling the Site for the purpose of producing training corpora.
Including Site content in training datasets.
Using Site content to fine-tune existing models or evaluate model performance.
Using Site content as part of retrieval-augmented generation (RAG) pipelines.
Any other AI / ML / LLM development purpose.

Technical implementation

The opt-out is implemented technically through robots.txt directives blocking known AI crawlers (GPTBot, ChatGPT-User, ClaudeBot, anthropic-ai, CCBot, Google-Extended, Bytespider, FacebookBot, Amazonbot, PerplexityBot, Diffbot). The list is not exhaustive; the prohibition is asserted against all AI training crawlers.

Important nuance for gaming sites: the AI search-result side (ChatGPT browsing, Perplexity, etc.) sometimes generates traffic for gaming queries. The opt-out we assert is for training; we do not necessarily block all AI-search query bots that retrieve information at request time without training on it. The robots.txt configuration is reviewed for this distinction.

5. Why this position

Catalog accuracy matters. AI-generated wrong attributions or genre tags propagate errors at scale across thousands of catalog entries.
Developer respect. The catalog respects developer attribution; AI-confabulated wrong credits are a form of disrespect to actual game creators.
AdSense compliance. AdSense's content-quality policies favor original, human-curated content over AI-generated content at scale.
Reader trust. Catalog integrity is the asset; AI-generated wrong content erodes trust over time.