Back to feed
OpenAI Blog·

Doppel’s AI defense system stops attacks before they spread

Signal
45
Hype
65
In three linesDoppel, an AI defense system using GPT-5 and reinforcement fine-tuning, detects and blocks deepfake and impersonation attacks before spread. Reduces analyst workload by 80% and response times from hours to minutes.
Read source
Your take?
GPTOpenAIAI safetyReinforcement learning

Summary generated by Claude — human-verified