According to @OpenAI, adversarial fine-tuning of the gpt-oss-120b model did not achieve high capability under their Preparedness Framework, even after robust adjustments. External experts reviewed OpenAI's methodology, indicating a move towards establishing new safety standards for open-weight AI models (source: openai.com). For crypto traders, this suggests that AI-driven trading algorithms based on open-weight models may present operational risks and may not deliver top-tier performance. This development could influence sentiment around AI-related crypto tokens and impact investment strategies in projects leveraging advanced AI for automated trading.