Using AI to Detect Rug Pulls — A Case Study

1. How this post-mortem works #

Selection logic: take recent rug-pull / exit-scam types already publicly confirmed by Etherscan / Solscan / DEXTools or multiple media outlets, and pick 3 representative shapes (illustrative templates below):

• Type A: DeFi protocol on Solana, TVL in the low single-digit millions at rug time
• Type B: meme-coin + GameFi concept on Base, market cap reaching roughly tens of millions
• Type C: infrastructure project on Ethereum mainnet, slow drain over several weeks, cumulative on the order of a million

Method: assemble all public information available in the 7 days before the rug into a single research pack for each AI. That pack includes:

Run each AI several times per project, take the consensus of the outputs, then match it against what actually happened.

2. Scenario A: caught — "ghost-deployer team" #

Type A is a DeFi protocol on Solana. In this kind of situation, steadier models (the Claude type) tend to flag high-risk in most runs. The reasoning typically hits three points:

Point 1: deployer address history. Feed Claude the deployer address and ask whether it had deployed any other contracts before this project. With Solscan data, it can flag that the address previously deployed several other contracts — some already abandoned with the owner pulling liquidity. This is what AI is genuinely good at: pattern matching across similar address histories.

Point 2: reverse-checking LinkedIn. If the project website lists team members with LinkedIn links, Perplexity can scrape those public profiles and often find that some show an unmistakable synthetic-profile pattern — very few connections, recently created, endorsements that look copy-pasted — and call it "likely synthetic identity team page." A human doing this cross-check would burn an hour or two; the AI gives a draft in minutes.

Point 3: TVL growth vs holder growth divergence. If Dune data shows TVL rising sharply over a short window while unique holder count barely grows, ChatGPT-class models will cut straight to it: "TVL growth is dominated by a handful of large addresses, not organic adoption. If they exit, you get a cascade."

This kind of Type A project tends to rug shortly after such a window. Here AI is genuinely useful — give it enough material and it parallel-processes several due-diligence checks in minutes that a human would spend an afternoon on.

3. Scenario B: caught — "copy-paste contract + fake team" #

Type B brands itself as "GameFi + meme" on Base. The core AI signal here is contract plagiarism.

Drop the verified source code into ChatGPT Code Interpreter and ask it to run a similarity comparison against known open-source token contracts. A common result: the contract is highly similar to the source of an earlier rug project, with only renamed variables and event names. High similarity by itself is not proof of a rug, but combined with "new-address deployer + fully anonymous team + recently registered Twitter account", the AI typically scores it high risk.

The other thing AI nails here: the whitepaper. Feed a multi-dozen-page whitepaper to Claude (one shot, long context) and it often flags several hard problems, for example:

• The Q2 roadmap claimed "partnership with X" — X had never acknowledged any partnership on Twitter.
• "Audited by CertiK" — but CertiK's public database has no matching record.
• The tokenomics chapter claims "team + advisors + private sale" sums to 67%, but the pie chart in the same document labels it 35%.
• A co-founder bio cited "former Coinbase engineer" — no GitHub commit history under that name at Coinbase's public repos.

The audit item is the killer. "Claims to be audited but cannot be found in the auditor's database" shows up in rug projects at an absurdly high frequency. The moment AI sees this, it pegs the risk high. This kind of "what the project claims vs what the public record says" cross-check is where AI is structurally strong — it has no emotional stake and is not seduced by the project's marketing.

4. Scenario C: missed — the slow rug #

This is the most valuable section of this article. Type C is an infrastructure project on Ethereum mainnet, with the rug playing out over several weeks and a cumulative drain on the order of a million dollars. In this kind of situation, the AI's runs tend to all score it "low" or "medium" risk.

Why did the AI miss it? Four reasons, in hindsight:

Reason 1: the contract was genuinely audited. This kind of project may have paid a second-tier firm for a real audit, with the report on GitHub and queryable in a public database. Once the AI sees "audited + report verifiable" it scores the contract-level rug risk low. But contract-level safety does not stop the team from slowly draining the treasury wallet via multisig — and that move is invisible at the contract layer.

Reason 2: the team was real-name, with real LinkedIn profiles. The team members may use real names, with LinkedIn profiles created years ago, high connection counts, and big-name previous employers. The AI scores "team verifiable" as a positive. But real-name teams can rug too — they just use language like "strategic pivot" and bleed the treasury over weeks instead of one violent withdrawal.

Reason 3: the rug signal lived in the governance forum, not on-chain. A retrospective dig through the Discord governance channel often shows that, weeks before the rug, the team had been seeding RFC posts hinting at "project direction adjustments". Those signals are text, buried in a large pile of governance posts. If the "social sentiment" you fed the AI was only an aggregated keyword summary — not raw discussion text — it can't catch second-order signals like "governance channel discussion density spiking + team response delays".

Reason 4: a slow rug has no breakpoint. AI (and most rug-detection tooling in general) leans on "anomaly events" as signals — TVL crashes, large transfers, sudden liquidity removal. A slow rug pulls a small amount per day over weeks; no single day looks unusual, and it only adds up to about a million. This is where rugs are evolving: away from violent one-shot exits, toward boiling the frog.

Conclusion from Type C: AI is good against old-school rugs (ghost deployer + copy-paste contract + fake identity). It is near-useless against slow rugs. That single lesson is the one most worth remembering when shaping a due-diligence process.

5. The 5 signal types AI is actually good at #

Combining the wins and the loss across all 3 cases, AI has a clear capability boundary on rug detection. These 5 signal categories are where it genuinely helps:

1. Deployer address history: what other contracts that address has deployed in the past 6-12 months and how they ended. This is the largest speed advantage AI has over a human — a human does this once in an hour, AI does it in minutes.
2. Contract source-code similarity: especially similarity against known rug contracts. ChatGPT Code Interpreter or Claude handle this well.
3. Synthetic identity verification: cross-checking LinkedIn / GitHub / posting history. Perplexity with web access is fastest here.
4. Whitepaper claims vs public record: do the claimed auditors / investors / partners actually appear in their respective public databases? Claude is strongest on long-document reading.
5. On-chain metric divergence: do TVL, holder count, and volume tell a consistent story? Dune data piped into ChatGPT or Claude.

What AI almost cannot do: smell the boiling-frog rug. Slow drains, governance manipulation, long-term value dilution — these require humans who have been tracking a project's culture over time.

6. A reusable prompt template #

You can run this template against every new project you look at. About half an hour of work blocks the majority of old-school rugs:

You are a strict crypto project due-diligence analyst. Based on the materials below, give me a rug-pull risk assessment (0-10 score).

Materials:
[1] Contract deployer address: {ADDRESS}
[2] Contract verified source: {SOURCE_CODE}
[3] Team page LinkedIn link list: {LINKS}
[4] Whitepaper PDF: {ATTACHED}
[5] Last 30 days of on-chain metrics: TVL={X}, Holders={Y}, Top10 concentration={Z}%
[6] Project's claimed auditor: {AUDITOR}

Requirements:
1. For every red-flag signal, give "evidence + rating (low/medium/high)". No hand-waving.
2. Do not predict whether it will rug. Describe the current risk structure only.
3. If the material is insufficient to judge an item, say "insufficient information".
4. End with a 0-10 score, plus the 3 strongest "do not invest" reasons (or write "no clear red flags" if there are none).
5. List 3 things you cannot see (your own blind spots).

That last item — "list 3 things you cannot see" — is worth adding deliberately. Forcing the AI to admit its limits makes the output more trustworthy, not less. When the AI explicitly says "I cannot see internal Discord discussions", you know that is the part a human has to cover manually.

AI does not block every rug. But it blocks about 70% of the low-effort ones, which frees your actual brainpower for deep tracking of the projects that look genuinely legit.

Open Binance Research → See the full Prompt Library →

— AI Trade Lab, 2026-04-25