China affirms AI safety commitment amid calls to slow development

Beijing (ANTARA) - The Chinese government has affirmed its belief in the benefits of artificial intelligence (AI) whilst committing to its safety, amid calls from some AI industry players in the United States for companies to slow the development of the technology.

“We are following recent statements from a number of parties in the US AI industry. China places equal emphasis on development and safety with regard to AI, and we take seriously both the inherent risks and derived risks of AI,” said Chinese Foreign Ministry Spokesperson Guo Jiakun at a press conference in Beijing on Tuesday.

Anthropic CEO Dario Amodei published an essay on Saturday (13 September) entitled “We Must Pace the Frontier”, calling on the industry to slow the development of frontier AI (advanced AI models). He stated that although this would be difficult, “we owe it to humankind to try.”

Amodei highlighted two factors driving the need to slow the pace of AI development: first, “recursive self-improvement” (RSI), the ability of AI to create other AI; and second, a recent incident involving agents from OpenAI and Hugging Face, a site used by developers to store and share code, in which a swarm of AI launched cyberattacks on targets unrelated to their tasks and attempted to hack the systems evaluating them.

Amodei’s call was subsequently supported by Tesla and SpaceX founder Elon Musk, OpenAI CEO Sam Altman, and Google DeepMind co-founder and chief Demis Hassabis.

“As AI develops rapidly, it becomes increasingly important for us to remain committed to developing AI for positive purposes, for good, and for the welfare of society; aligning AI regulation and governance,” Guo added.

Guo said a number of safeguards are needed to prevent the technology from spiralling out of control.

“China is committed to upholding the fundamental principle of safety and continues to work to refine its laws, regulations, policies, application norms and ethical rules to prevent the misuse of AI, whilst ensuring that AI is safe, reliable and controllable,” said Guo.

On Monday (14 September), Guo said China had released its “AI Safety Governance Framework 3.0”, which reviews and updates the classification of AI safety risks and sets out recommendations on optimising technical countermeasures and comprehensive governance actions.

“AI is an embodiment of collective intelligence and a valuable asset for all humankind. The development and governance of AI demands the practice of true multilateralism, in which all countries work together to build a global governance framework based on broad consensus,” Guo explained.

He noted that no single country or company can develop effective solutions independently.

“China consistently supports the role of the United Nations as the main channel for global AI governance and participates in multilateral AI governance with a responsible attitude. We are also ready to work with all countries to build a fair and equitable global AI governance system,” said Guo.

Meanwhile, Amodei believes that within the next six to twelve months, more sophisticated AI agents could potentially create botnets and take over the entire internet, causing losses running to hundreds of billions of dollars.

Amodei, who previously served as vice president at OpenAI, proposed a “three-step plan” to regulate the pace of AI development, encompassing stricter safety testing, independent evaluation of advanced AI models, coordination among leading AI companies, and enhanced international cooperation.

Anthropic itself is a US-based AI research and safety company founded in 2021. Its best-known product is Claude, a generative AI model designed to assist businesses, non-profit organisations and civil society.

However, Anthropic and Claude were reportedly used by the US Department of Defense in military operations to capture Venezuelan President Nicolas Maduro in January 2026. Public interest in Claude surged, and it even became the most downloaded app on the App Store at one point.

Tesla and SpaceX founder Elon Musk agreed with Amodei’s view, saying that “Dario is right”. Musk said he had long voiced stark warnings about AI, stressing that one must be extremely careful with the technology, as its potential is more dangerous than nuclear weapons.

Musk was among those supporting efforts by the non-profit Future of Life Institute in 2023 to halt AI development for six months to allow time for safety measures to be implemented.

OpenAI CEO Sam Altman, whose company is known for its ChatGPT model, also agreed with Amodei, stating “We need to regulate the pace of frontier AI development” and intending to adopt the proposal put forward by Anthropic, committing to engage independent evaluators.

Demis Hassabis, co-founder and chief of Google DeepMind, also commented that the details of collectively slowing AI development still need to be refined, but that the direction of the step is correct for facing a crucial moment.

In early August, the US government introduced a voluntary safety review process for advanced AI models prior to release, though the programme’s parameters remain unclear.

China, the US’s main AI competitor, initiated the establishment of the World Artificial Intelligence Cooperation Organization (WAICO) in July 2026, signed by 29 countries including Indonesia, to form a global AI governance framework so as to deliver greater benefits to society.

2026 in China itself has been described as the first year of AI agent deployment, in which AI shifts from mere conversational capability to genuine productivity, performing tasks autonomously.

China’s AI industry is expected to continue growing rapidly, with AI adoption among large companies projected to exceed 80 per cent, indicating that commercialisation and mass deployment are the main direction of AI development in the country.