Microsoft will conduct security ratings for the models provided to cloud customers.

Microsoft to Provide Security Ratings for Cloud Customers' Models

Microsoft will start ranking artificial intelligence models based on their security performance. As part of selling AI products from companies such as OpenAI and Elon Musk's xAI to cloud customers, Microsoft hopes to build customer trust in this way. Sarah Bird, Microsoft's head of responsible artificial intelligence, said the company will soon add a "Security" category to its "Model Leaderboard". The leaderboard is a feature launched by Microsoft for developers this month to rank model iterations from multiple providers, including China's DeepSeek and France's Mistral. The leaderboard is accessible to tens of thousands of customers using the Azure Foundry developer platform and is expected to influence which AI models and applications customers purchase through Microsoft. Microsoft currently ranks three metrics: quality, cost, and throughput.

—— Financial Times

via Xiangfengqi Reference Express - Telegram Channel