AI Race Accelerates Despite Global Warnings as Self-Improving Systems Move Closer to Reality
WASHINGTON: Growing international calls to slow the development of artificial intelligence are facing increasing challenges as new advances bring the technology closer to creating and improving future versions of itself with minimal human intervention.
The debate intensified following a recent report from Anthropic, which revealed that more than 80 per cent of the code used to power its Claude AI chatbot is now generated by the system itself. The figure marks a dramatic increase from early 2025, when only a small fraction of the code was self-generated.
According to Anthropic’s AI development framework, current AI systems have reached the fourth of five developmental stages. At this level, models can independently execute programming tasks, generate software code and assign work to other digital systems with limited human oversight.
The company projects that the next stage could involve autonomous AI agents capable of designing, training and improving their own successor models. Such a development would represent a major milestone in artificial intelligence, potentially allowing future systems to evolve continuously without direct human involvement.
The rapid pace of progress has renewed concerns among researchers, technology experts and policymakers who have long warned about the risks associated with increasingly powerful AI systems.
One of the most notable warnings came in 2023 when the Future of Life Institute published an open letter signed by prominent figures, including Elon Musk, urging a temporary pause in the development of AI systems more advanced than GPT-4.
The signatories warned that highly capable AI could contribute to widespread job displacement, the spread of misinformation, cybersecurity threats and the possibility of humans losing meaningful control over critical systems.
Concerns have also extended beyond the technology sector. In May, Pope Leo XIV called for stronger international oversight of artificial intelligence, urging governments to prevent the technology from becoming a tool in geopolitical and military competition.
Despite these warnings, investment in AI continues to surge. Industry analysts project global spending on artificial intelligence will reach approximately US$2.59 trillion in 2026, representing a 47 per cent increase from the previous year. Investment is expected to climb further to US$3.49 trillion in 2027 as businesses and governments race to adopt the technology.
As AI capabilities continue to expand, experts say the challenge facing policymakers will be balancing innovation and economic growth with safeguards designed to ensure the technology remains transparent, accountable and under human control.