OpenAI has published its latest threat report examining how malicious actors combine AI models with websites and social platforms for coordinated attacks. The February 2026 report builds on two years of research into AI abuse patterns, revealing that threat actors rarely limit themselves to single platforms or models. Instead, they orchestrate campaigns across multiple channels.

According to OpenAI's findings, threat actors integrate AI models into broader operational workflows alongside traditional tools like social media accounts and websites. The report includes case studies showing how groups like Chinese influence operators leverage different AI models at various stages of their campaigns. This multi-platform approach makes detection and defense more complex for security teams.

OpenAI emphasizes that understanding these patterns helps the industry and society better identify and mitigate AI-enabled threats. The company continues publishing these reports to share insights about evolving attack methodologies, building on previous work examining prompt injection attacks and data safety measures. This transparency aims to strengthen collective defenses against increasingly sophisticated AI abuse.