Most acceptable-use policies are long because they reserve the right to act on anything. This one is short because it does not.
The whole list
Both are removed the moment we know, without a court order and without waiting for one. Neither has anything to do with copyright, and no complaint is needed to trigger either.
Refused 1 of 2
Child sexual abuse material
Any depiction of a minor in a sexual context, hosted, cached, linked or distributed. There is no threshold, no artistic exception and no jurisdictional argument that changes this.
Removed on sight · reported · account closed · no appeal, no refund
Refused 2 of 2
Material organising violence against people
Direct incitement, target lists, and doxxing published to get someone hurt. Discussing violence is not organising it — we are describing operational material, not opinion.
Removed on sight · reported · account closed · no appeal, no refund
That is the entire content policy. Everything below this line describes things we restrict because they break other people’s machines — a different question, with a different answer, and never a reason to close an account on a first offence.
Not grounds for anything
Named rather than implied, because “legitimate use” in someone else’s policy is a word that means whatever they need it to mean the day they need it.
If your workload is not on this list and you want certainty before you pay, ask support and you will get a straight yes or no in writing — which you can then hold us to.
A different question
A host that claims to permit literally everything is not telling the truth, because nobody lets attacks leave their own network. Here is exactly where the line is, and why it is there.
Attacks launched from your server
Denial of service, credential stuffing, mass exploitation. Not a moral position: outbound attacks get our prefixes null-routed by other people, which takes down customers who did nothing.
Port closed, ticket opened, 24 h to fix it
Unsolicited bulk mail
Mail to people who never asked gets the whole range blacklisted within hours, and the range is shared. Bulk mail to a list you own is fine and always has been.
Port 25 rate-limited, then closed on repeat
Spoofed source addresses
Traffic leaving our network with an address that is not yours is the raw material of reflection attacks. Egress filtering is on by default and is not removable.
Filtered at the edge, silently
Consuming a shared resource to exhaustion
On a shared-core plan, saturating a host degrades every neighbour. Sustained load is fine; starving the hypervisor is not. Reserved cores exist precisely so this stops being a question.
Throttled, then a suggestion to move plan
The operative text. It says the same thing as the rest of this page, in the form a lawyer would want to read it in. Where the two disagree, this section governs.
This policy binds every account holder, and everything running on any service we provide — virtual servers, dedicated servers, addresses and ports. You are responsible for what runs on your service whether you put it there or someone you gave access to did.
It does not bind your users. If you run a platform, we look at you, not at them, and we expect you to have your own rules. We do not ask to see them.
Two categories of content are prohibited, without exception and without regard to where you or your users are:
These are removed the moment we are aware of them. We do not require a court order, a takedown notice or a complaint in the correct format, and we will not wait for one. The account is closed, the matter is reported to the appropriate authority, and no refund is issued for the remainder of the term.
This list has two items. It has never had a third, and adding one would be a material change to this policy, announced 30 days in advance under section 08.
Separately from content, the following damage infrastructure that is not yours, and are restricted for that reason alone:
Our first step in every one of these cases is a ticket and a deadline, not a suspension. We suspend only when the traffic is actively harming a third party and you are unreachable.
For the avoidance of doubt, none of the following is a breach of this policy, and none of them will cause your service to be suspended, throttled or reported:
We do not maintain a private list of “high-risk” customers, we do not price by risk category, and we do not have a category that gets watched more closely than another.
Complaints that are not about the two prohibited categories are logged, timestamped and forwarded to you verbatim within 24 hours. We do not evaluate their merits, and nothing on your service changes as a result of receiving one. What you do about it is yours to decide — including nothing.
The DMCA policy describes the copyright case specifically, and the transparency report publishes how many arrive.
Suspension decisions under this policy are made by a named engineer, not by an automated classifier and not by a first-line agent following a script. There is no system that scans customer disks or traffic; the only way we learn about content is that someone tells us.
Outside the two prohibited categories, you are contacted before anything is restricted, and you get a deadline that reflects the severity — 24 hours for active harm to a third party, longer for anything else.
A suspension stops the service; it does not delete it. Your data remains for 14 days, during which you can take a copy. Termination for the two prohibited categories is immediate and permanent, and data is destroyed rather than retained.
Credit for a suspension we later determine was unjustified is the remainder of the term plus one month, applied without asking.
A material change — anything that would make a currently permitted use impermissible — is announced by email at least 30 days before it takes effect, and the previous version stays reachable at a stable address. If you do not accept a change, cancel before it takes effect and unused time is refunded pro rata in the coin you paid with.
Governed by the law of the Republic of Moldova. Reports of the two prohibited categories are read every day of the year, including weekends, and acted on before anything else.
Version 1.3 · last substantive change 2026-07-01 · in force since 2026-07-22. Material changes are announced by email 30 days before they take effect, and the previous version stays reachable.
Before you move in
If your case is not covered here, ask before you pay rather than after. A written answer from support is binding on us.
Open a ticketAn engineer, on the material itself, and the test is operational rather than moral: does this publication help a specific person get hurt. A manifesto is not a target list. A forum arguing about a war is not incitement. If a case is genuinely ambiguous we contact you before we act, and in the four years this policy has existed the ambiguous cases have been rare enough to count.
No. There is no scanner, no hash-matching, no traffic classifier and no sampling. The only way we ever learn what is on a service is that a human being reports it. That is a deliberate design decision rather than a gap, and it is the reason the privacy policy can state that we hold no traffic content — a scanner would make that sentence false the day it was switched on.
There is no number. There is no threshold at which volume converts into a suspension, and there is no internal counter per account that we watch. Notices are forwarded to you and that is the end of our involvement — the totals we publish in the transparency report are network-wide and are not broken down by customer.
No, and a host that promises it is promising something outside its control — Moldovan law can change, and we would follow it. What we can promise is the procedure: any addition is a material change, announced 30 days ahead by email, with the old version left reachable and a pro-rata refund for anyone who does not accept it. You would never learn about it from a suspension.
Read the rest
What happens when a complaint arrives is written down too, and so is the count of how many did.
Read next
Where to go from here
