OpenAI新研究：o1增加推理时间就能防攻击，网友：DeepSeek也受益

This finding may also benefit other models such as DeepSeek, indicating a broader impact on AI robustness.

Key Points

• OpenAI's Scaling Law improves the robustness of reasoning models against attacks.
• Extending reasoning time for model o1 significantly reduces vulnerability.
• DeepSeek may also benefit from the findings of this research.
• The study highlights the importance of reasoning duration in AI security.