Anthropic, an Artificial Intelligence company, has called for a coordinated global mechanism that would allow leading AI developers to slow down or temporarily pause the development of advanced AI systems if risks begin to outpace society’s ability to manage them.
The San Francisco-based company, known for its Claude family of AI models, warned that the rapid advancement of frontier AI could eventually lead to systems capable of improving themselves with limited human oversight.
Anthropic said such developments could arrive sooner than many governments, institutions, and regulators are prepared for, Reuters reported.
In a recent report and accompanying policy proposal, Anthropic argued that the world should have the option to temporarily slow AI progress to allow safety research, governance frameworks, and societal institutions to keep pace with technological advances.
The company stressed that any pause would need to be coordinated internationally, as unilateral action by a single company would likely prove ineffective if competitors continued developing increasingly powerful systems.
Anthropic’s proposal includes the creation of a globally verifiable framework that could monitor compliance among major AI developers, drawing comparisons to international agreements used in areas such as nuclear governance.
The company said its newly launched research initiatives will focus on studying the economic, social, and security implications of advanced AI systems.
The warning comes at a time when AI capabilities are increasing across the industry.
Anthropic revealed that AI tools are already playing a significant role in software development within the company, which highlights that advanced systems are contributing to the creation of new technologies.
Anthropic believes governments, researchers, and technology firms must work together to establish safety mechanisms before AI capabilities advance beyond existing oversight structures.
Join BusinessDay whatsapp Channel, to stay up to date
Open In Whatsapp
