Hackers Are Using ChatGPT Custom GPTs to Steal Crypto: The Attack Nobody Warned You About
Cybercriminals are now hiding inside ChatGPT itself, using OpenAI's custom GPT feature to run sophisticated social engineering attacks designed to drain crypto wallets.
The attack vector is called ClickFix, and it is spreading fast. Hackers build convincing custom GPTs that mimic legitimate crypto tools, support bots, or DeFi protocol assistants. Once a user interacts, the fake GPT delivers manipulated instructions that trick victims into executing malicious scripts on their own machines, handing over wallet credentials, seed phrases, or private keys without ever triggering a traditional phishing alarm.
Why This Attack Is Different
Most phishing attacks leave fingerprints. Fake URLs, suspicious sender addresses, broken English. ClickFix through custom GPTs leaves almost none of them.
The conversation feels natural because it is happening inside a trusted, verified platform. OpenAI's interface lends the attacker instant credibility. A user asking a "DeFi assistant GPT" how to connect their wallet has no obvious reason to be suspicious. By the time the malicious payload runs, the damage is already done.
Security researchers flagging the exploit note that the technique bypasses most conventional endpoint detection tools because the victim is essentially instructed to compromise their own system. There is no malware dropped in the traditional sense. The user clicks, pastes, or approves the action themselves.
The Bigger Problem: Ad Verification and Platform Accountability
This attack also exposes a serious gap in how AI platforms vet the tools published on their marketplaces. OpenAI's GPT Store allows third-party builders to publish custom assistants with minimal friction. That same low barrier to entry that made it a powerful tool for developers is now a doorway for bad actors.
The crypto industry has seen this movie before. Fake browser extensions, malicious DeFi front-ends, counterfeit wallet apps. Each time a new trusted platform emerges, attackers colonize it before security catches up. ChatGPT is simply the latest frontier.
Stricter ad verification and mandatory security audits for GPTs that reference financial or crypto topics are now being called for across the security research community. Whether OpenAI moves fast enough is an open question.
What Crypto Holders Should Do Right Now
Treat every custom GPT that asks you to copy, paste, or run anything on your device as a red flag, regardless of how polished it looks. Legitimate DeFi protocols do not deliver support through ChatGPT bots.
Verify any GPT's origin independently before engaging. Never enter seed phrases or private keys into any AI interface under any circumstances. Hardware wallet users should double-check that no script execution is being prompted on the connected device.
The sophistication of this attack will increase. Watching for new ClickFix variants targeting crypto users across AI platforms is now a baseline security habit, not an optional one.