Did Amazon CEO's Concerns Lead to Anthropic's Model Shutdown?

Behind AI startup Anthropic's sudden shutdown of access to two AI models to users worldwide on Friday, it has emerged that Andy Jassy, CEO of parent company Amazon, may have raised safety concerns.

According to reports, Jassy is said to have pointed out security and safety-related issues with some of the models provided by Anthropic within the company. Multiple sources believe that such concerns likely led directly to the decision to suddenly cut off access.

Anthropic restricted access to specific two models on a global scale, but has not publicly disclosed the detailed reasons. However, this move coincides with the U.S. government's efforts to strengthen AI regulation, and there is a growing view within the industry that regulatory pressure and internal governance issues at the company acted in combination.

Amazon announced a large-scale investment in Anthropic in the fall of 2023 and agreed to invest up to $4 billion (approximately 600 billion yen). Since then, the relationship between the two companies has gone beyond simple capital partnership, with technological integration between Amazon's cloud service AWS and Anthropic deepening. Against the backdrop of such close relations, it is reasonable to consider that Jassy's remarks could have influenced Anthropic's management decisions.

Discussions about AI safety have been rising rapidly in recent years, and calls for major AI companies to assess the risks of their own models and ensure transparency have been strengthening from both government and civil society. Anthropic's response in this case is attracting attention as an example of a company responding swiftly to such external pressure.

On the other hand, information is limited regarding what specific risks the two models targeted for access shutdown are believed to have, and when the shutdown will be lifted. Both Anthropic and Amazon spokespeople have refrained from commenting, and the clarification of the detailed circumstances is expected to take time. As the AI industry continues to expand rapidly, if CEO-level concerns at major tech companies have led directly to model shutdowns, it would be a case that offers important implications for the future direction of AI governance.