china taps models
| | |

China taps US AI models to train homegrown rivals as chip curbs bite

BEIJING, China – Chinese entities are adopting increasingly sophisticated ways to use output from leading American AI models to train domestic rivals, documents show, as China aims to broaden its AI strategy and offset tightening US curbs on its access to chips.

A Reuters review of more than 300 Chinese academic papers, patents, corporate disclosures, recruitment notices and government documents reveals how the practice — known as distillation — has expanded beyond academic experiments into mainstream industrial use in China.

The records describe Chinese researchers using US models to produce answers to questions, build computer code, and find solutions to difficult problems, with Chinese companies then developing software to automate the process and specialist workers preparing material for training smaller domestic models.

The documents also show that Beijing is encouraging distillation to make AI cheaper and easier to deploy across China’s powerful manufacturing sector.

Distillation is not illegal and is prevalent in other parts of the world. While the process cannot reproduce a leading model’s full capabilities or eliminate the need for advanced computing power, it can help developers preserve selected skills in smaller systems that cost less to operate, potentially reducing their reliance on American technology.

China’s broadening use of distillation could undercut the effectiveness of US curbs on the sale of advanced chips to China, as its developers can build AI capabilities by drawing on those of their American rivals. Washington said on Monday it would further restrict China’s access to such chips from companies like Nvidia.

Distillation is also a source of friction between the world’s two biggest economies ahead of the November 18-19 Asia-Pacific Economic Cooperation summit in Shenzhen, where US President Donald Trump and his Chinese counterpart Xi Jinping are expected to meet.

US authorities publicly accused Chinese companies in September of aggressive, targeted distillation of American AI models on an industrial scale.

And at meetings in Chengdu in July, US Commerce Department officials raised concerns with Chinese regulators about developers collecting output from leading American AI models, warning that some activities could amount to intellectual-property theft, three people familiar with the discussions said.

Chinese officials defended distillation as a standard AI-development technique and said they were open to acting where the conduct constituted IP theft, but made no concrete commitments, according to a US official familiar with the previously unreported exchange.

The US Commerce Department and China’s Cyberspace Administration did not respond to requests for comment.

From American answers to Chinese models

Distillation involves a powerful ‘teacher’ model generating answers or demonstrating how to complete tasks, which developers use to train a smaller ‘student’ system.

For Chinese developers facing constraints on computing power, the attraction is in making one expensive query do repeated work, according to a former Alibaba Group engineer who spoke on condition of anonymity. An American model can generate the answer to a difficult question once, allowing a Chinese team to retain useful examples and train a local system around them.

In a November 2024 paper, Shanghai Jiao Tong University researchers described selecting about 5,000 answers from OpenAI’s o1 model, translating them using GPT-4o mini and assembling 10,750 bilingual training examples to further train Alibaba’s Qwen2.5-72B model.

The researchers also cautioned that excessive reliance on distillation could stifle original innovation, leaving developers dependent on breakthroughs made by more advanced model providers.

The project illustrated how American model output can become reusable data, allowing developers to improve domestic systems without repeatedly consulting the original provider.

Alibaba researchers described a more systematic approach in an April 2025 paper, using OpenAI’s GPT-4 and GPT-4o to generate training material in languages other than Chinese, then refining and scoring responses before training smaller Qwen models.

In a February 2026 paper, researchers from Shanghai Jiao Tong University and Chinese food-delivery company Meituan described using GPT-4o to generate plans for coordinating AI agents working together on programming problems, helping train a smaller model to organise such tasks.

Shanghai Jiao Tong University, Alibaba and Meituan did not respond to requests for comment.

A growing geopolitical dispute

Among leading American AI companies, OpenAI’s terms restrict using its output to develop competing models, subject to exceptions, while Anthropic similarly restricts training competing systems without permission.

A US official familiar with the July Chengdu regulatory discussions said Washington believes industrial-scale distillation is the main reason Chinese AI models have narrowed the gap with their American rivals.

Two people familiar with Chinese developers’ operations disputed that assessment, saying distillation was only one factor alongside advances in engineering, algorithms, data and model design.

Chinese government documents do not instruct companies to distil American models but encourage distillation as part of Beijing’s push to spread AI across the world’s second-largest economy.

A national AI+Manufacturing policy package published by eight Chinese government agencies in January includes guidance supporting model training, fine-tuning and distillation for industrial applications. Shanghai and Shenzhen have also encouraged techniques to reduce computing requirements.

The policies reflect Beijing’s ambition to deploy smaller, cheaper AI systems across industries.

Concerned by distillation’s growing geopolitical sensitivities, Chinese AI developers have in recent months asked Beijing’s industry regulator for guidance on the boundaries between ordinary use of foreign models and organised collection of output for training, according to one person familiar with the discussions.

China’s Ministry of Industry and Information Technology did not respond to a request for comment. – Rappler.com

Similar Posts

  • | |

    West Indies take first-innings lead 

      West Indies all-rounder Justin Greaves produced a sensational bowling performance, claiming five wickets for just 27 runs in 11 overs, to put his team in complete control after dismissing Pakistan for 282 in the first innings. His remarkable spell dismantled Pakistan’s lower order and ensured the hosts secured a 29-run first-innings lead, giving West Indies a significant advantage in the Test match. Greaves was well supported by the rest of the bowling attack. Veteran fast bowler Kemar Roach and pacer Shamar Joseph each picked up two wickets, while Jayden Seales contributed with one dismissal. Together, the West Indies bowlers maintained relentless pressure throughout Pakistan’s innings and prevented the visitors from posting a commanding total. Earlier in the day, Pakistan resumed their innings at 199 for 3, with captain Shan Masood and vice-captain Salman Ali Agha at the crease. The pair displayed patience and composure, frustrating the West Indies bowlers as they added 47 valuable runs for the fourth wicket. Their partnership helped Pakistan build a solid platform and raised hopes of a substantial first-innings total. Shan Masood continued his impressive form and reached a memorable century. The left-handed batter scored 109 runs from 184 deliveries, striking 12 boundaries during his well-crafted innings. It was the seventh Test century of his career and his first Test hundred against the West Indies. Masood’s innings was marked by discipline, determination, and excellent shot selection as he anchored Pakistan’s batting effort for most of the innings.   However, soon after Masood’s dismissal, Pakistan’s batting lineup suffered a dramatic collapse. Salman Ali Agha, who had provided useful support with 20 runs off 60 balls, was bowled by Shamar Joseph, triggering the downfall of the middle and lower order.   Justin Greaves then produced a devastating spell that completely changed the course of the innings. He first dismissed Aamir Jamal for 3 runs, followed by the wicket of debutant Ali Usman, who was out without scoring. Greaves then removed Mohammad Rizwan for 12 runs, leaving Pakistan in serious trouble. His disciplined line and length exposed the visitors’ lower-order weaknesses and prevented any meaningful recovery.   Pakistan eventually lost their remaining wickets in quick succession and were bowled out for 282. Despite Shan Masood’s excellent century, the visitors failed to capitalize on the strong start they had made earlier in the day, with the lower order unable to offer much resistance against the disciplined West Indies bowling attack. By securing a 29-run first-innings lead, West Indies gained an important psychological and tactical advantage in the match. The lead places the home side in a strong position as they prepare to bat in their second innings, while Pakistan will need an outstanding bowling performance to make a comeback. With momentum firmly on the hosts’ side, the contest has shifted in West Indies’ favor, setting up an intriguing battle in the remaining days of the Test match.

  • |

    Diplomatic storm erupts over podcast banter involving Japanese prime minister

    Australian Prime Minister Anthony Albanese is facing intense political backlash after a podcast appearance triggered a high-profile diplomatic row with Tokyo. The controversy stems from an episode, recorded in July, in which Albanese discussed a pair of prized crown melons presented to him during a May bilateral meeting by Japanese Prime Minister Sanae Takaichi. During the segment, podcast host Nikki Osborne made an explicit joke comparing the fruit to female anatomy, to which Albanese responded with agreement. The exchange went largely unnoticed upon release but reignited publicly after Shingo Yamagami, Japan’s former ambassador to Australia, published a scathing opinion piece accusing the Australian leader of casually mocking a key strategic ally. Yamagami described the remarks as a sexist insult that betrayed the close diplomatic bond between the two nations, noting that crown melons are highly valued cultural gifts in Japan, often worth upwards of 30,000 yen (£150) each. Opposition leaders in Canberra immediately seized on the commentary, demanding an official apology from Albanese for what they termed “classless” and disrespectful conduct. Albanese rejected calls to apologise in parliament, insisting that the partnership between Australia and Japan remains as strong as ever and pointing out that the joke was initiated by the host rather than himself. Senior Labor Party ministers also defended the prime minister, arguing that the comments were taken out of context by political opponents and former diplomats. The episode marks the second time Albanese has faced criticism over his appearance on the show, following a prior apology for crude remarks regarding pop star Kylie Minogue.

Leave a Reply

Your email address will not be published. Required fields are marked *