Acceptable use policy
Last updated 10 September 2026
This policy sets out what you may not do with a AIHostBox service. It exists because the software we host — agents that browse, automations that send, models that generate — can be misused in ways ordinary web hosting cannot.
1. Unlawful and harmful use
- Anything illegal in the United States or in your jurisdiction.
- Child sexual abuse material, in any form, generated or otherwise. Accounts are terminated immediately and reported.
- Harassment, stalking, doxxing, or content intended to threaten or intimidate.
- Malware distribution, intrusion attempts, credential stuffing, or scanning networks you do not own.
2. AI-specific restrictions
- Impersonation. Do not use voice cloning, text generation or chat automation to impersonate a real person or organisation in order to deceive.
- Bulk unsolicited messaging. Automation platforms may not be used to send spam over email, SMS or messaging networks, or to operate engagement farms.
- Scraping. Web agents and crawlers must respect robots directives and rate limits, and may not be used to bypass anti-bot protection or to collect personal data at scale.
- Fraud and manipulation. No generated reviews, fake identities, synthetic engagement or content designed to manipulate markets.
- Automated decisions about people — hiring, credit, housing, insurance — require a human in the loop. Do not deploy a model as the sole decision-maker over someone’s life.
3. Resource use
Plans are sized for the application they run. Cryptocurrency mining, distributed computing projects, torrent seeding and public open proxies are not permitted. Sustained saturation of a shared host beyond your plan’s allocation may lead us to ask you to upgrade.
4. Open relays and open endpoints
Do not run an unauthenticated model endpoint, MCP server or automation webhook that is open to the public internet. Beyond the abuse risk it is a direct route to your own API keys being spent by strangers.
5. Reporting and enforcement
Report abuse to [email protected]. Depending on severity we may contact you, suspend the service, or terminate it without refund. We prefer the first, and in practice most issues are a misconfiguration rather than intent.