Editorial standards & trust

We cover a young field that is easy to hype. The following four are hard constraints, not aspirations:

1. Fact, inference and opinion are kept separate——What a third party can independently verify is fact. What we derive from facts but cannot yet confirm is inference, and is labelled as such. Our own position is always marked as an OceanAlt take, presented separately from the facts rather than blended into the narrative.
2. No numbers without a source——Any transaction volume, user count, funding amount or market size must trace to a publicly accessible source. If it can't, we don't publish it — or we say the data isn't available. We would rather leave a gap on the page than fill it with a plausible-looking number.
3. Subjects have a right of reply——For critical judgements about a named organization, we seek comment before publishing. If they decline or don't respond within a reasonable window, we say so in the piece rather than omitting the step.
4. Translation is not authorship——Republished or adapted material must credit and link the original. Our Chinese and English editions are the same newsroom expressing the same facts twice — not unreviewed machine translation.

Requirements vary by content type, but these three always apply:

• Key facts need two independent sources. Where only one exists, the piece must say so explicitly.
• Primary beats secondary: official docs, on-chain data, regulatory notices and company filings outrank press summaries. Anything verifiable on-chain gets a clickable explorer link.
• Timestamped: policy and protocol versions move fast, so any claim about a current state carries the date it was checked.

Required structure by content type:

When we find an error we fix it — but never silently. Any change that affects how a reader understands the piece is logged at the bottom of the article and aggregated here. Three kinds:

• Correction — A factual error. The original wording is preserved or the change is described explicitly.
• Clarification — The facts were right but the wording could mislead.
• Update — New information emerged after publication; the original wasn't wrong but needs continuation.

To report an issue, use the contact form — we'll verify it and publish the outcome.

Published corrections

No corrections published yet. That doesn't mean we won't get things wrong — when we do, the record will appear here, and it won't be removed.

Our research comes in two kinds, with different methods and different labels:

• Desk research — Analysis based on public documents, on-chain data, regulatory texts and company disclosures. Limitations — small samples, inconsistent definitions, incomplete disclosure — are stated at the end.
• Hands-on experiments — We build systems, break them ourselves, and publish the process. Every episode of our Attack Lab series broke our own implementation, then fixed and retested it. Conclusions apply only to what we tested and are not extrapolated to the industry.

The definitions, sources and refresh cadence behind our homepage metrics have their own page: Sources & method。 For ratings, see the RAP framework。

We use AI in producing content and don't hide it. The boundaries:

• AI does: source collection, first drafts, Chinese–English translation, structural editing.
• Humans do: topic selection, fact-checking, conclusions, and the decision to publish. Nothing goes live that a human hasn't reviewed.
• AI is not permitted to: invent data, invent quotes, or invent source links. Any such error is handled as a correction and logged publicly.

Everything on this site is for information only and is not investment, legal or compliance advice. Regulation of agentic payments and stablecoins is still moving in most jurisdictions; for actual business decisions, consult a licensed professional in your jurisdiction.

About our own products, to be clear about where things stand: our compliance gateway currently runs 11 gates (9 always on); on-chain settlement runs on the Base Sepolia testnet and has not been connected to mainnet funds; Tron screening is sanctions-list only, without on-chain heuristics; and we have not undergone a third-party security audit. All of this is on our proof page.

OceanAlt accepts commercial partnerships and commissioned assessments, but commercial relationships do not influence our rating conclusions; all such relationships are disclosed on the relevant rating pages.

In practice:

• Paid content is labelled as sponsored, is visually distinguishable from editorial, and never appears in our research or ratings sections.
• Commissioned assessments use the same methodology as public ratings. A client may dispute facts; a client may not request a different conclusion.
• Where we have any commercial, investment or personnel relationship with a rated entity, that relationship is disclosed on the page. If it can't be disclosed, we don't rate them.
• As of today, OceanAlt has no outside investors and no commercial agreement with any rated entity. This sentence will be updated as that changes.