The Chinese AI firm used its rare public appearance to deliver an unusually sober warning about the risks of rapid AI adoption, in stark contrast to the triumphant tone that followed its model’s shock debut earlier this year.

Speaking at the World Internet Conference in Wuzhen, Zhejiang province, DeepSeek’s first major public engagement in nearly a year, senior researcher Chen Deli argued that AI could reshape labour markets far faster than governments and societies are prepared for, predicting large-scale disruption over the coming decade.

“AI’s short-term impact may appear positive,” Chen told delegates, including Chinese policymakers and executives from leading technology firms. “But in five to 10 years, we may see significant job losses. Over ten to twenty years, the risks to human society could be considerable.”

The comments reflect a shift in tone for DeepSeek, which rose to international prominence in January when its R1 model was reported to match or exceed performance of leading Western systems on certain reasoning tests, despite being trained at a fraction of the cost typically associated with cutting-edge AI development.

A peer-reviewed paper later said the R1 system was trained for approximately US$294,000 using 512 Nvidia H800 chips, an unusually efficient approach compared with the multi-million-dollar budgets associated with US labs such as OpenAI and Google DeepMind. The release triggered market volatility in US semiconductor stocks and fuelled debate over whether China was closing the technological gap faster than expected.

Founded in 2023 by Liang Wenfeng, who previously co-founded the quantitative hedge fund High-Flyer, DeepSeek has positioned itself at the forefront of China’s domestic AI ambitions. The company has also distinguished itself by adopting a more open model-weight philosophy than many global rivals, prompting praise from open-source advocates and scrutiny from governments concerned about data governance, security and content controls.

Since its breakthrough, DeepSeek has largely maintained a low public profile. However, it continued releasing upgrades, including a V3.1 system in mid-2025 that promised improved reasoning speed and hybrid inference capabilities.

In September, it also entered a partnership with Huawei to support compliance-focused model development aimed at increasing the reliability of content moderation, a move closely watched in the context of China’s strict information-control regime.

The company’s reappearance comes at a pivotal moment in the global AI race. Western policymakers have warned that China’s advances could undermine the impact of export controls designed to slow Beijing’s access to advanced semiconductor technology. Meanwhile, international regulators are intensifying scrutiny of both AI safety and potential strategic misuse.

Chen’s stark tone in Wuzhen suggests DeepSeek is keen to frame itself not only as a technical competitor to Western firms, but also as an early voice cautioning against unrestrained deployment.

While he emphasised AI’s benefits, Chen concluded that governments and businesses must “prepare responsibly” for social transformation. His remarks may resonate beyond China, as world leaders continue to grapple with how to regulate AI ecosystems that are evolving faster than existing governance frameworks.

DeepSeek did not make new product announcements at the event, but its return to public view, and the warning it delivered, ensured that its influence on the global debate remains firmly intact.

RELATED STORIES

China’s DeepSeek AI sparks global tech shift, challenges US Big Tech dominance

OpenAI predicts major AI breakthroughs by 2028, despite ‘serious weaknesses’

AI data centre boom strains power grids and cooling supply chains: report