Skip to main content

AI-washing, onchain-washing and token-washing

Project cards on DecAIHub sometimes include red flag labels: AI-washing, onchain-washing, or token-washing. This page explains what they mean and how to read them.

In short: washing is a situation where a project claims a property (AI, onchain, token utility) but no verifiable evidence of that property has been found. This is not an accusation — it is a record of the gap between claims and artifacts.

A washing flag is not a verdict and not an accusation

A washing flag in the card means exactly one thing:

The project claims X, but as of the verification date we did not find Tier-1 evidence of X.

The reason can be entirely mundane:

  • Documentation exists, but we did not find it — it may have been published after the verification date, placed in a non-obvious location, or available only upon request.
  • The project is at an early stage — the product is in development, artifacts have not yet appeared, and the roadmap has not yet been realized.
  • The project uses a closed model — code, metrics, or architecture are not published for business reasons, even though the technology genuinely exists.
  • Information exists but not at the Tier-1 level — the project is described in interviews or on an aggregator (Tier-2), but official documentation or code has not been published.
  • We simply did not find it — when reviewing hundreds of projects, something can be missed.

DecAIHub does not draw conclusions about a project's intentions. We do not know why evidence is absent — and we do not speculate. The card records the state as of the verification date (Last verified), and when new artifacts appear, the assessment is revised.

AI-washing

What it is. A project uses the words "AI," "machine learning," or "neural network" in its marketing, but:

  • no description of what exactly the AI does has been found;
  • no demos, metrics, or technical documentation have been found;
  • no public code related to the AI component has been found;
  • based on available sources, the AI component is limited to calling a third-party API (e.g., OpenAI), with no proprietary processing logic detected.

How it appears in the card. ai_score = 0 or ai_score = 1 (declaration of intent only), while AI messaging is actively used in marketing.

Tier-1 artifacts that reduce this risk:

  • a public repository with AI component code and release history;
  • a technical description of the model/architecture with version and date;
  • quality metrics with context (what was measured, on what data, when);
  • demos or working examples available for verification.

Onchain-washing

What it is. A project positions itself as "decentralized" or "onchain," but:

  • contract addresses have not been found, or the contracts found contain no logic beyond the token;
  • based on available sources, the blockchain is used primarily for token issuance;
  • based on available sources, significant logic runs on a conventional server;
  • replacing the blockchain with a conventional database, based on available data, does not change the product's key properties.

How it appears in the card. onchain_score = 0–1, while the project calls itself a "decentralized" or "onchain protocol."

Tier-1 artifacts that reduce this risk:

  • contract addresses on a public network with verifiable transactions;
  • documentation describing which rules and states live onchain;
  • a rationale for why blockchain is critical (the "Replace-the-Chain" test);
  • a smart contract audit.

Token-washing

What it is. A project has a token, but:

  • based on available sources, it is not clear what the token pays for or to whom;
  • based on available sources, the token does not participate in the product's operation;
  • based on available sources, the token is replaceable with USDC, a subscription, or another mechanism without loss of functionality;
  • utility is described in generic terms ("for the ecosystem," "for the community"), with no specific functions found.

How it appears in the card. token_score = 0–1, while the token is actively promoted.

Tier-1 artifacts that reduce this risk:

  • documentation with specific token functions (payment, staking, slashing, access);
  • on-chain mechanics where the token participates in protocol logic;
  • a description of what breaks if the token is removed;
  • transparent tokenomics with vesting, distribution, and treasury.

Common anti-patterns

Anti-pattern Washing type What to look for
"Has a repository," but as of the verification date it is empty or unrelated to the product AI-washing Presence of releases, tests, CI, alignment with the whitepaper
"Has an audit," but no public report found Onchain-washing Link to a specific report with contract addresses
"Has AI metrics," but without methodology and version AI-washing Context: what was measured, on what data, when
"Onchain," but no contract addresses Onchain-washing Explorer, addresses, transactions
"Token for the ecosystem" with no specific functions Token-washing Documentation with token functions in the protocol
"Decentralized AI" with no verifiable artifacts AI + onchain washing Tier-1 evidence for both directions

See also

No comments to display

Back to top