Amazon Breaks Silence on AI Safety Debate: Models Must Be Tested Before Release

Amazon has publicly responded for the first time to the intensifying AI safety debate, stating that AI models will only be released after rigorous testing to ensure safety. The company has not followed calls from some industry leaders to slow down AI development, warning instead that AI advancement carries risks if not pursued jointly by the industry.

The statement, made by an Amazon spokesperson on September 17, 2026, comes days after several major AI labs urged a more cautious pace. Amazon's position draws attention due to its multiple roles in the AI supply chain: it develops its own frontier models such as the Nova series through AWS and its AGI division, serves as a key infrastructure provider for many AI labs, and offers both its own and competitors' models through its Bedrock platform. Amazon is also a major investor in Anthropic, one of the companies that called for slowing AI development, having invested $13 billion as of April and committing up to an additional $20 billion if certain commercial targets are met.

The industry remains divided. On September 12, Anthropic CEO Dario Amodei published a lengthy post urging slower AI development, a stance supported by OpenAI, xAI, Google DeepMind, and Microsoft. However, Nvidia and Meta CEOs said on September 15 that AI risks should be assessed and controlled by individual companies. President Trump dismissed AI safety concerns as a "scam" and is expected to meet with AI executives this week or early next week.