OpenAI’s latest threat report reveals how bad actors are abusing AI models—and what they’re doing to stop it. Over the past two years, the company has tracked a pattern: most threats don’t rely solely on AI but mix it with traditional tools like fake websites and social media. One case study even shows a Chinese influence operator hopping between different AI models during their workflow, proving attacks can span multiple platforms. The report emphasizes that AI misuse is rarely isolated, and sharing these insights helps the whole industry spot and block these schemes faster. Why it matters: As AI becomes more accessible, understanding these attack patterns is crucial for staying ahead of evolving threats—not just for companies like OpenAI, but for everyone who relies on these tools.