Acceptable use

The public statement of the same rules that already run against every broadcast before its search is allowed to start. Nothing here is a policy we intend to apply; it is a description of code that is running.

These are refused automatically, before the search runs. Nothing is charged, because nothing happened.

• Phishing and scams — anything built to get a payment or a credential out of somebody under a false pretence.
• Illegal solicitation.
• Hate speech — attacks on people for what they are.
• Sexual content.
• Financial fraud patterns — guaranteed returns, recruitment pyramids, and the rest of the family, in any category, not only investment.

Every post in investment, legal financial, medical fundraiser and relief is read by a person before it sends, whoever sends it. These are the categories where the risk is the claim rather than the reach — a false one costs somebody money, health or a legal right — so no automatic check stands in for a person. Payment does not substitute.

Elsewhere, a post the automatic check has doubts about but does not block may send. A recipient's report of any post reaches a person.

• Do not work around moderation, tier limits or rate limits, including by opening a second account to escape a restriction on the first.
• Do not try to identify a subscriber, or reach them anywhere other than the channel they agreed to.
• Do not put false information in a verification request.
• Do not scrape or harvest beyond what the product hands you directly.
• Do not advertise a business in the dating category. It exists for people describing themselves, and a matchmaking service posting there is using a channel built on the assumption that a person is on the other end.

• A blocked broadcast simply does not run. No charge, no strike — though trying repeatedly is itself conduct, and reads as such.
• A new sender's first post is checked by the AI and then by a person before it sends. Once it is approved, later posts are checked by the AI only and go out straight away.
• 2 or more spam complaints and a rate above 0.3% over 30 days suspend the account, automatically. There is no smaller second chance: every sender's mail leaves from the same address, so one sender's complaints reach everybody's inbox.
• A confirmed violation of the two sections above can end the account. It does not take your unused balance with it — see the refunds page.

Those are the real numbers the code checks, not illustrative ones.

Every delivered message carries a way to report it, and every public alert in Explore does too. A report on a public alert removes it from Explore immediately, before anybody reviews it — the review decides whether it goes back, not whether it comes down.

Reports on delivered broadcasts reach the same moderation queue that reviews flagged content before it sends.

These rules are the ones in the code. When the code changes, this page changes in the same commit, and active accounts are told before a material change takes effect.