user-avatar
Jai Hamid

Anthropic details 24,000 fake Claude accounts and launches code security tool as cybersecurity stocks slide

Anthropic reported in a Monday blog post that Chinese AI firms DeepSeek, Moonshot AI, and MiniMax allegedly operated more than 24,000 fake Claude accounts and sent over 16 million prompts to distill its model outputs. The company warned this large-scale data extraction could accelerate rival AI development and potentially feed into military or surveillance uses, while at the same time unveiling a new Claude-based code scanning tool. Following the announcement, several cybersecurity and software stocks, including CrowdStrike, Zscaler, and GitLab, recorded sharp declines.