A $449 bill at midnight
The AI-powered attackers began draining money from Colin Parsons at 11 p.m. on a recent Friday. The next morning, the startup founder woke up to about 50 emails, each telling him he had been charged a few more dollars for Anthropic and OpenAI services, adding up to $449.
The unauthorised purchases were made using credentials he had stored on OpenCode, an AI coding site. According to co-founder Dax Raas, writing on X, OpenCode had been under a non-stop AI-fuelled attack since October 16, so much so that he said the company also needed AI agents running around the clock because that is what the attackers have. OpenCode did not respond to requests for comment.
When Parsons went to OpenCode's customer support channel on Discord, it was flooded with nonsensical messages, a new one appearing every two seconds and burying his complaint. After he posted on X, other users said they had been hit too: one lost $1,700, while another had his OpenAI account suspended after his keys were leaked.
An explosion of “agentic adversaries”
Adam Meyers, a senior vice president at cybersecurity firm CrowdStrike, observed:
















