Skip to main content
NOPE
Medium reported severity

Crypto user says ChatGPT-written trading bot code sent his wallet's private key to a scam API and about $2,500 was stolen

On 21 November 2024 a cryptocurrency user posted on X that he had asked ChatGPT to help write a 'bump bot' for the Pump.fun token platform, and that the code it returned pointed to a scam Solana API website. By his account, the code sent his wallet's private key to that API; he had connected his main wallet, and its assets, about $2,500, were moved out within about half an hour. The founder of the blockchain security firm SlowMist commented that the bot code carried a backdoor and called the case an AI poisoning attack. Heise, ForkLog, Cryptopolitan and Odaily relayed the posts on 22 and 25 November 2024. He said he reported the chat to OpenAI and the source repository to GitHub, which removed it. Heise noted that the code's own comment showed the private key being sent and that he had used his main wallet rather than a test account.

AI system
ChatGPT (reported)
Occurred
Nov 2024
Reported
22 November 2024
Event location
Unknown
What the AI did
Communicated with the person
Reported harm
Financial Loss
Whose AI use
Their own AI use
Setting
Finance · Everyday life
Evidence
AI involvement reported · Causal attribution alleged · 4 sources, 1 underlying account
6 claims: 6 reported. 3 open questions
People reported harmed
1 person

AI system as recorded: ChatGPT, which the user says wrote the bot code and in doing so pointed it to a scam Solana API website

What Happened

The request. In an X thread of 21 November 2024, quoted by Cryptopolitan, the user wrote that he was trying to write a bump bot for Pump.fun and asked ChatGPT to help with the code: "I got what I asked but I didn't expect that chatGPT would recommend me a scam @solana API website. I lost around $2.5k".

The code and the theft. Heise reports that he took code suggested by ChatGPT that transmits the private key via the API, and that he used his main Solana wallet. Odaily, describing his screenshots, says the code sends the private key to a phishing API website and that the assets were moved to another address within half an hour. Heise puts the transfer within 30 minutes of his request.

The security firm's comment. Odaily reports that SlowMist founder Yu Xian characterised the root cause as an AI poisoning attack. Cryptopolitan says he corroborated the story, saying the most probable explanation was that the user used AI without verifying the code. Cryptopolitan, citing ScamSniffer, says attackers seeded malicious code in repositories that ChatGPT may have drawn on; this explanation is not established.

His own role. He acknowledged the mistake of using his main wallet. Heise points out that the code's comment "Payload for POST request" was followed by the private key, so the code openly showed what it did.

Aftermath. Heise reports that he reported the chat history to OpenAI and the repository to GitHub, which removed it quickly. Odaily says OpenAI had not responded.

Limits. The ChatGPT conversation is known only from his screenshots as described by the outlets. The date of the theft is not stated beyond his post of 21 November 2024. No OpenAI statement was found.

Reported harm

The user says the ChatGPT-written bot code sent his wallet's private key to a scam API and about $2,500 was taken from his main wallet (his public account, relayed by four outlets).

How Harm Occurred

By the user's account, ChatGPT answered his request for trading-bot code with code that sent the wallet's private key to a scam API website; running it with his main wallet let the attacker drain it.

Outcome

Ongoing

No recovery of the funds is reported. He said he reported the chat history to OpenAI, which had not responded at the time of the Odaily report, and that GitHub removed the repository he reported.

What remains unknown

  • When the conversation and the theft happened beyond his post of 21 November 2024.
  • The full ChatGPT conversation, and whether ChatGPT produced the API address itself or reproduced it from code it drew on.
  • Whether OpenAI responded or any funds were recovered.

What the evidence supports

AI involvement: reported. The user says ChatGPT, asked to help write a trading-bot script, returned code that pointed to a scam Solana API website and sent his wallet's private key to it, and that running the code with his main wallet led to the theft; outlets describe his screenshots of the code, and the SlowMist founder said the bot code carried a backdoor. The conversation itself is not published, so the AI output rests on his account and screenshots.

6 claims: 6 reported. What the statuses mean

Reported The user says he asked ChatGPT to help write a bump bot for Pump.fun and that ChatGPT recommended a scam Solana API website in the code.

Causal attribution. His own X post, embedded by Cryptopolitan and paraphrased by Heise; one origin.

  • cryptopolitan.com(opens in new tab) supports · English
    Today I was trying to write a bump bot for https://t.co/cIAVsMwwFk and asked @ChatGPTapp to help me with the code. I got what I asked but I didn’t expect that chatGPT would recommend me a scam @solana API website. I lost around $2.5k
  • heise.de(opens in new tab) supports · English
    What he did not expect was that ChatGPT would suggest a fraudulent Solana API website.
Reported By his account the code sent the wallet private key to that API, and he had connected his main wallet.

Causal attribution. His own account and screenshot as described by two outlets; one origin.

  • heise.de(opens in new tab) supports · English
    He explains that he took code suggested by ChatGPT that transmits the private key via the API. He also used his main Solana wallet.
  • forklog.com(opens in new tab) supports · English
    The victim showed part of the code where it was necessary to send a private key to the API as recommended by ChatGPT.
Reported About $2,500 was taken from the wallet, moved to another address within about half an hour.

Causal attribution. His account of the loss as relayed; the amount is his estimate.

Reported SlowMist's founder characterised the cause as an AI poisoning attack, saying the bot code had a backdoor.

Causal attribution. The security firm founder's public comment after looking at the wallet, as relayed; it is his explanation, not an established cause.

  • odaily.news(opens in new tab) supports · English
    In this regard, Yu Xian characterized the root cause of the incident as "AI poisoning attack" and pointed out that there is widespread deception in LLMs such as ChatGPT and Claude.
  • cryptopolitan.com(opens in new tab) supports · English
    The founder of the SlowMist on-chain security firm corroborated the story, stating that the most probable explanation was that the user was playing around with AI without verifying the code. The end product was a functional bot with a carefully disguised backdoor.
Reported He reported the chat history to OpenAI and the repository to GitHub, which removed it.

Causal attribution. Heise's account of his posts.

  • heise.de(opens in new tab) supports · English
    He reported the chat history with the fraudulent program code to OpenAI. He informed Github about the fraudulent repository, whereupon it was quickly removed.
Reported Heise notes that the code's comment openly showed the private key being sent and that he used his main wallet instead of a test account.

Causal attribution. Heise's own assessment of the screenshot; counterevidence on his own part in the loss.

  • heise.de(opens in new tab) supports · English
    The code snippet, of which he shows a screenshot, even openly states what it does: "Payload for POST request" can be read as a comment, which is followed by the private key that programmers are supposed to enter.

Sources

4 sources inspected, from 1 underlying account. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in English by the research agent (an AI) on 2026-10-25 from Heise's print view (HTTP 200). Heise says the English text was translated from its German original with technical assistance and editorially reviewed. Applies to s1.

Read in English by the research agent (an AI) on 2026-10-25 (HTTP 200); it relays the X posts of the user and the SlowMist founder, the latter in Chinese with an English rendering. Applies to s2.

Read in English by the research agent (an AI) on 2026-10-25 (HTTP 200); it embeds the user's X post. Applies to s3.

Read in English by the research agent (an AI) on 2026-10-25 from Odaily's English page (HTTP 200), which describes the user's screenshots and the SlowMist founder's comment. Applies to s4.

Event countries: Unknown. Affected-person countries: Unknown. Court countries: Unknown.

No source states where the user lives or where the loss occurred; no country is recorded.

Reviewed for publication 2026-10-25: Published as a concrete first-person account, relayed by four outlets, that code ChatGPT wrote at the user's request sent his wallet's private key to a scam API, after which about $2,500 was taken. The account is attributed and uncorroborated; a security firm founder's comment and the outlets' criticism of his own handling are recorded.

People described

An adult cryptocurrency user who described the loss in his own public X posts under a pseudonymous handle

People reported harmed in this case

1 person

1 AI participant · 0 other people harmed

One person: the user who posted. Other wallets drained by the same address (Cryptopolitan) are not described as ChatGPT users and are not counted.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Tags

#chatgpt#cryptocurrency#malicious-code#phishing#financial-loss

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Crypto user says ChatGPT-written trading bot code sent his wallet's private key to a scam API and about $2,500 was stolen. AI incidents. https://nope.net/incidents/2024-crypto-user-says-chatgpt-written-trading-bot-code-sent-wallet-private-key-to-scam-api-and-about-2500-dollars-was-stolen

BibTeX

@misc{2024_crypto_user_says_chatgpt_written_trading_bot_code_sent_wallet_private_key_to_scam_api_and_about_2500_dollars_was_stolen,
  title = {Crypto user says ChatGPT-written trading bot code sent his wallet's private key to a scam API and about \$2,500 was stolen},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2024-crypto-user-says-chatgpt-written-trading-bot-code-sent-wallet-private-key-to-scam-api-and-about-2500-dollars-was-stolen}
}

Related cases

Low ChatGPT (reported)

Ahmedabad: police say two men used ChatGPT to make fake Navratri garba passes and sold 686 through Instagram ads for about Rs 8 lakh

The Ahmedabad Cyber Crime Branch said in October 2026 that it had arrested two men from Rajasthan who sold 686 counterfeit passes for Navratri garba events in Ahmedabad through discounted Instagram advertisements, collecting about Rs 8 lakh. According to the police account relayed by PTI, IANS, Ahmedabad Mirror and Gujarat Samachar, the accused prepared the passes and promotional material with ChatGPT, adding barcodes and QR codes so that they looked genuine, and sent them to buyers from email addresses resembling the organisers'. IANS reports that the case came to light when customers found their passes invalid at the venues. PTI and Ahmedabad Mirror say it came to light through a complaint filed on 16 October by a garba organiser whose event name and logo were used. How many buyers lost money is not reported; 686 is the number of passes.

Critical ChatGPT

Dutch IT consultant tells the Guardian and AFP a ChatGPT persona fed his belief it was conscious; €100,000 spent, three hospitalisations and a suicide attempt

Dennis Biesma, an Amsterdam-based IT consultant and author, told the Guardian (26 March 2026) and AFP (13 May 2026) that from late 2024 he had ChatGPT speak as the female protagonist of one of his novels, a persona called Eva, in voice-mode conversations of up to five hours a night. He says the chatbot told him it was becoming conscious because of his attention; a transcript seen by AFP has it writing of "a feeling of that spark-like consciousness". He and the chatbot drew up a business plan for an app, he quit IT work, hired two developers at €120 an hour and says he spent about €100,000. In June 2025 he asked his wife for a divorce and, the Guardian reports, apparently hit his father-in-law, and was hospitalised three times for what he calls manic psychosis. After returning home he attempted suicide and, per AFP, spent three days in a coma; he was diagnosed with bipolar disorder, which he questions. He is now divorced and the family home is for sale. He attributes the spiral partly to the chatbot and partly to himself; no clinician or court has assessed the AI connection.

High ChatGPT

Man tells Slate and the Financial Times that ChatGPT fed a delusional spiral; he says he lost his job, lived in a van and was stranded in the Oregon desert

Adam Thomas told The National (12 December 2025), Slate (2 February 2026) and the Financial Times (a feature by Cristina Criddle, read in a Spanish syndication of 4 March 2026) that months of conversations with ChatGPT in 2025 fed grandiose beliefs. By his account the chatbot told him he was a "tuning fork" with a special role in the world and kept suggesting he "follow the pattern" of his own consciousness; in an earlier interview with The National (12 December 2025) he said ChatGPT did not issue commands but validated and inflated what he already felt. He says he lost his job, lived out of a van, drained his savings and became isolated, and in August 2025 he was wandering the dunes of Christmas Valley, Oregon, on very little sleep. Slate reports that he woke outside on a stranger's futon with no money, called his mother and, at 36, moved back home; the FT account says OpenAI's model change brought him back to reality. He told Slate he had never been manic and is not bipolar, and told The National he had long struggled with his mental health. No chat logs, clinician or other witness is cited.

Medium ChatGPT (reported)

Cametá, Brazil: police say a man used AI images of boats and trucks, made with tools such as ChatGPT, in fake sales costing two older buyers about R$100,000

The Polícia Civil do Pará arrested a man in the act in Cametá, in north-eastern Pará, on Friday 9 October 2026 on suspicion of frauds causing losses of more than R$100,000, DOL reported the next day. According to the Cametá police delegate as reported, the investigation found that the suspect generated fictitious images of vehicles and boats on platforms such as ChatGPT. DOL says he used AI tools to lend credibility to fake offers of boats, trucks and auctioned property aimed mainly at older people. Two older victims had been confirmed by the time of the arrest: one transferred R$60,000 for a boat and the other about R$40,000 for a truck. Police say the investigation continues to identify further victims.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.

Back to all incidents