WiretechnologyBrief
Viral ChatGPT Memory Prompt Shows a Big Risk for Crypto Holders
A viral prompt shows what ChatGPT memory knows about users. Crypto holders face added phishing risk, while OpenAI offers ways to switch it off.
Brief
A viral prompt has users upload a picture of an empty gray room and ask ChatGPT to fill it with what the chatbot knows about them. The furnishings draw on past conversations, and some people found the results unsettling. According to OpenAI, ChatGPT memory has two parts: saved memories that record details such as a name, preferences and goals, and chat history reference, available to Plus and Pro subscribers, which lets the model use older conversations even when nothing was saved. Because public posts can expose a profile to strangers, crypto holders face extra risk. Scammers could use stored details about holdings or exchanges to write more convincing phishing messages. Users can view, delete or turn off memory in the Personalization settings, and Temporary Chat neither reads nor updates it.
Why it matters: It shows how much personal data an AI chatbot can retain, which matters to crypto holders whose wallet and exchange details could aid phishing.
Read the full story
Dextape context
Built from Dextape’s own wire, stories and glossary.
Dextape story on this event
Terms in this story
- Phishing
- Fake sites, messages or support agents that trick you into revealing a seed phrase or signing a malicious transaction.
Mentioned in this story
Open a coin page for its current price.