Hong Kong crypto firm HashKey moves into AI amid drive for more blockchain automation

HashKey’s first AI product is designed for content moderation, but it envisions a future of AI agents trading on blockchain on behalf of users

Reading Time:2 minutes

Hong Kong-based cryptocurrency exchange operator HashKey is joining the artificial intelligence race with a product designed for content moderation. Photo: Shutterstock

Published: 4:30pm, 3 Apr 2025

Hong Kong-based HashKey Group, a licensed cryptocurrency exchange operator, is making its foray into the artificial intelligence (AI) market with the launch of its first generative AI product, underscoring the rising interest from Web3 firms in combining AI and blockchain technologies.

ModAI, unveiled on Thursday, is designed to automate content moderation, generate humanlike engagement, and offer customer support. The tool could “slash operational costs by 80 per cent and double user activity”, the company said in a statement.

ModAI uses a combination of large language models (LLMs) from DeepSeek, OpenAI, Anthropic and Meta Platforms, according to HashKey. It currently supports Chinese and English, with Japanese and Vietnamese, among other languages, to be added later this year, the company said.

The AI initiative marks a new business vertical for HashKey, which began exploring new areas of business two years ago. It initially pursued training its own LLM – the technology underpinning chatbots such as ChatGPT – but scrapped the plan due to high costs, Sean Wang, HashKey’s vice-president of product who leads the AI team, told the Post during an interview on Thursday.

Recent innovations from Chinese start-up DeepSeek have reignited those efforts, according to Wang. DeepSeek’s cost-efficient V3 and R1 models have significantly lowered the barriers to LLM training, he said.

The company may use distillation – an approach popularised by DeepSeek for training smaller models to mimic the behaviour of larger ones – to build LLMs tailored for the Web3 community.

Wang said the company expected AI agents, autonomous programs capable of performing tasks on behalf of users, to become a key application of LLMs. These agents could, for example, automatically invest in virtual assets based on a user’s capital and target returns, he explained.