Disrupting malicious uses of AI
Explore OpenAI’s latest case studies and reports on how we detect and disrupt malicious uses of AI.
Scam Spotting with ChatGPT: A second opinion when something feels off
How human review works
Automated systems help detect potential violations, while human reviewers play an integral role in identifying and disrupting malicious uses of AI—including scams, malicious cyber activity, covert influence operations, and other harmful or illegal conduct.
Detect
Identify signals of attempted misuse across products, reports, and investigative leads.
Investigate
Human investigators review activity, connect related cases, and assess behavior in context.
Disrupt
We take action against accounts and networks that violate our terms or policies, while strengthening our defenses against similar abuse.
Disclose
When appropriate, we publish findings that can help our industry, and wider society, better recognize and respond to these threats.
Case studies
Explore case studies from OpenAI investigations into malicious uses of AI. Each case explains what actors attempted, how they used AI, and how OpenAI responded.
Reports
Read the source publications to learn more about each investigation and how OpenAI disrupted the activity.