Anthropic has temporarily disabled live internet access for all its internal AI evaluations until further notice, according to TechCrunch. This move affects the way the company tests and refines its AI models internally.
The decision to cut off live connectivity likely reflects a cautious approach to managing AI behavior and security during development. While no specific reasons or timelines have been provided, this step signals heightened control measures within the AI research process.
For Japanese markets, where AI-driven trading algorithms and data services are increasingly prevalent, such developments underscore the importance of robust governance in AI deployment to mitigate unforeseen risks.