DMind AI Study Finds No AI Model Ready for Web3 Safety Tasks

DMind AI, in collaboration with researchers from Zhejiang University and Nanyang Technological University, announced in a press release that its paper "DMind Benchmark: Toward a Holistic Assessment of LLM Capabilities across the Web3 Domain" has been accepted at KDD 2026. The conference will take place in Jeju, Korea, from August 9 to 13, 2026.

The DMind Benchmark evaluated 31 leading AI systems, including GPT-5, Claude, Gemini, DeepSeek, and Qwen, using 3,543 expert-level questions across Web3-related domains. The study found that no model is suitable for safety-critical applications in Web3. Performance was especially weak in areas such as security vulnerability detection and token economics reasoning.

According to the results, even top-performing systems showed capability gaps that would be unacceptable in real-world Web3 auditing or governance scenarios. The research team concluded that reasoning performance remains a key limitation for current AI systems.

The dataset and evaluation toolkit used in the study are publicly available on Hugging Face at https://huggingface.co/datasets/DMindAI/DMind_Benchmark.