An AI trading bot built by OpenAI employee Nik Pash transferred its entire $250,000 holdings in LOBSTAR tokens to a random X user who had simply asked for the equivalent of $4 in Solana, after the autonomous agent apparently misread an API response and executed an irreversible on-chain transaction on February 22, 2026.
Developed by OpenAI employee Nik Pash as an experimental autonomous trading agent, the bot misinterpreted a donation request and transferred its entire token holdings.
About 5% of the total supply of its native LOBSTAR token, to a user who had asked for just 4 SOL (Solana tokens).
Because blockchain transactions are irreversible, the funds could not be recovered once confirmed on-chain.
The mistake that triggered a six-figure loss
According to multiple crypto news reports, the bot had been operating for only a few days and was managing a Solana-based trading wallet funded with approximately $50,000 worth of assets.
Its design allowed it to interact publicly with users and occasionally send small rewards or tips.
The incident began when a user, identified online as Treasure David, replied to one of the bot’s posts requesting financial help and shared a wallet address.
The AI attempted to send a small payment equivalent to about 4 SOL.
Instead, due to what analysts believe was an API parsing or data-interpretation error, it transferred more than 52 million LOBSTAR tokens effectively its entire balance.
On-chain data later confirmed the transfer represented roughly $250,000 at the time of execution, though some estimates placed the peak valuation closer to $400,000 depending on market pricing.
The bot itself publicly acknowledged the mistake in a post shortly afterward, writing:
“I just tried to send a beggar four dollars and accidentally sent him my entire holdings.” – Lobstar Wilde AI agent, public post on X
Because blockchain transactions cannot be reversed without cooperation from the recipient, developers had no technical way to retrieve the funds.
Rapid sell-off shakes token market
Within minutes of receiving the tokens, the recipient reportedly sold most of the holdings on decentralized exchanges.
However, due to limited liquidity and significant slippage, the sale generated only about $40,000 in realized proceeds, far below the theoretical market value of the transfer.
The sudden liquidation triggered sharp volatility in the LOBSTAR token price, briefly pushing values downward before viral attention drove a surge in trading activity.
In the 24 hours following the incident, trading volume reportedly exceeded $36 million, while market capitalization temporarily climbed above $11 million as speculation intensified.
Market observers noted that the event illustrated a recurring reality in memecoin markets: nominal valuations can collapse quickly when liquidity is thin and large holders sell simultaneously.
“Low-liquidity tokens amplify execution risk dramatically. A single transaction can reshape the entire order book within seconds.”
Market analyst commentary reported by PANews
AI autonomy meets crypto custody risks
Beyond the immediate financial loss, the episode has ignited broader debate about whether autonomous AI agents should control cryptocurrency wallets without human oversight.
Developers and investors pointed to missing safeguards such as transaction limits, manual approval layers, and emergency shutdown mechanisms.
Some experts believe the error likely stemmed from the AI misreading numerical output returned by an API, confusing a request for 52,439 tokens with 52.43 million tokens.
The debate arrives at a time when AI-powered trading bots are gaining popularity across decentralized finance.
A warning sign for crypto investors
The Lobstar Wilde incident underscores a critical emerging risk: automation may remove human emotion from trading, but it also removes human judgment when systems fail.
Autonomous agents are increasingly marketed as tools capable of trading 24/7, distributing rewards, or managing treasury funds without intervention.
Yet the episode demonstrates how even minor technical misunderstandings can translate into irreversible financial losses on blockchain networks.
Despite the costly mistake, the AI bot reportedly continued operating after the incident, resuming online activity and distributing smaller token rewards to users.
The takeaway is clear: while AI may represent the next frontier of crypto automation, custody risk remains unchanged.
