KawaiiGPT, a free malicious large language model (LLM) first spotted in July 2025 and now at version 2.5, empowers novice cybercriminals with tools for phishing emails, ransomware notes, and attack scripts, drastically lowering the entry barrier for cybercrime.