Attackers are leveraging the chat-sharing features of ChatGPT and Claude to distribute malware disguised as error messages or installation guides. This tactic allows the malware to evade security measures by being hosted on reputable domains, raising concerns about the safety of shared AI conversations.