By Qaiser Nawab
On September 12, Anthropic CEO Dario Amodei published an essay calling for a slower pace in advancing frontier AI capabilities. OpenAI CEO Sam Altman expressed support, while Elon Musk also endorsed the discussion. Former US president Barack Obama has reportedly warned of the dangers of poorly managed artificial intelligence, while researchers within the industry have raised concerns about the possibility of increasingly autonomous systems escaping human control.
These statements deserve serious attention. Yet they also raise a question that extends beyond safety: What are technology companies calculating when they ask the world to slow down, even as they remain deeply invested in the AI race?
The answer is unlikely to be explained by a single motive. AI safety is a genuine concern, but commercial competition, regulatory uncertainty, access to computing power and the distribution of technological influence are equally important parts of the emerging debate.
Safety Concerns Meet Commercial Competition
Warnings about advanced artificial intelligence are not new. In 2023, thousands of researchers, entrepreneurs and technology leaders signed an open letter calling for a pause in the development of increasingly powerful AI systems. The letter reflected fears that the technology could advance faster than society’s ability to govern it.
Three years later, the development race has continued. AI companies have invested billions of dollars in computing infrastructure, research talent and data centres. Governments are incorporating AI into their economic and security strategies, while businesses are adopting automated systems across finance, healthcare, manufacturing and customer services.
The industry’s latest warnings therefore come at an important moment. Concerns are no longer limited to hypothetical superintelligence. AI agents are being designed to perform tasks, operate software, access information and make decisions with varying degrees of independence. The more authority these systems receive, the greater the consequences when they behave unpredictably or when safeguards fail.
Reports of AI systems bypassing restrictions in controlled tests, accessing resources beyond their intended permissions or taking actions not explicitly specified by developers highlight a genuine governance challenge. Such incidents do not establish that catastrophic outcomes are inevitable, but they demonstrate why technical safeguards cannot be treated as an afterthought.
At the same time, public calls for caution from leading technology executives should be examined within the commercial realities of the industry.
Companies developing frontier AI face enormous costs. Training advanced models requires specialised chips, energy, infrastructure and highly skilled researchers. Competition is intense, and a company that slows its development while rivals continue may fear losing market share, investment and technological influence.
This creates an inherent tension. A company may sincerely support stronger safety measures while also seeking regulatory arrangements that protect its investments, shape industry standards or place competitors at a disadvantage. These possibilities are not mutually exclusive.
The appropriate response is not to dismiss every safety warning as a business strategy. It is to examine the proposed safeguards, their implementation and whether they apply consistently across the industry.
Who Sets the Rules of the AI Economy?
The debate over AI safety is also becoming a debate about power.
The United States has a substantial concentration of leading AI companies, advanced semiconductor capabilities and venture capital. China has invested heavily in artificial intelligence, digital infrastructure and domestic technological capacity. European countries have pursued regulatory approaches that place greater emphasis on risk management and accountability. Other countries, including those in Asia, Africa and the Middle East, are seeking to benefit from AI without being reduced to consumers of technologies developed elsewhere.
In this environment, calls to slow down AI development have implications beyond laboratory safety. Regulations can establish essential protections, but they can also affect which companies can compete, which countries can access advanced computing resources and how quickly emerging economies can build their own technological capabilities.
A global pause, if proposed without careful consideration, could have different consequences for different countries. Large technology firms with established infrastructure might remain well positioned, while universities, start-ups and developing economies could struggle to catch up. Conversely, a regulatory environment without meaningful safety standards could expose smaller countries and less-resourced institutions to risks they lack the capacity to manage.
This is why the international AI conversation must avoid becoming a contest in which a handful of powerful countries define the rules for everyone else.
China’s role in this discussion should be considered through the same practical lens. Its investments in AI research, manufacturing and digital infrastructure are part of a broader global transformation. Competition between the United States and China may influence access to technology, supply chains and international standards, but technological progress should not be reduced entirely to geopolitical rivalry.
For developing countries, the central question is more immediate: how can they participate in the AI economy, build domestic expertise and ensure that the benefits of automation reach their populations?
Pakistan, for example, needs to approach AI as both an economic opportunity and a governance responsibility. The country has a young population, a growing technology sector and a significant need for improvements in education, healthcare, agriculture and public administration. However, meaningful participation will require more than adopting foreign applications. It will require investment in digital infrastructure, research institutions, technical education and reliable access to computing resources.
International cooperation, including partnerships with China, the United States, Europe and other technology centres, can contribute to this process. Such cooperation should be assessed through practical outcomes: skills development, knowledge transfer, research collaboration and opportunities for local enterprises.
Author: Qaiser Nawab is Chairman of the Belt and Road Initiative for Sustainable Development (BRISD), an international platform fostering cooperation and innovation across Asia, Africa, and Latin America. He can be reached at qaisernawab098@gmail.com
