Microsoft CEO Satya Nadella warns of AI risks

The rapid development of Artificial Intelligence (AI) technologies is bringing not only new opportunities but also unexpected risks. In his latest statement, Microsoft CEO Satya Nadella issued an unexpected and candid warning to the tech world. He believes that companies using services from major labs like OpenAI and Anthropic are unknowingly putting the future of their businesses at risk. This is reported by Techcrunch.com reports .

Nadella emphasized that today's AI model users are essentially "paying twice." First, they pay directly for AI tokens and services, and second, and most importantly, they are providing their most confidential and valuable business data to these models. This data is used to make the model smarter, which can result in large tech labs learning their clients' internal secrets and becoming their competitors.

"Trojan Horse" effect and data theft

According to ixbt.com, many investors and experts in Silicon Valley are comparing these models to a "Trojan Horse." The more data companies upload to AI models to increase their work efficiency, the more these models learn. Satya Nadella called this process "intellectual property leakage." According to him, the model learns from users' corrections, queries, and tools used by agents, absorbing the company's unique know-how.

This situation is particularly relevant for startups and large enterprises in developing markets like Uzbekistan. Local companies may inadvertently expose their customer bases, strategic plans, and internal algorithms when using global AI services. Nadella described this type of knowledge as "wealth that a competitor could never buy, but AI models are getting for free."

Distillation and the issue of fair use

The Microsoft executive raised another important issue: companies creating AI models use open data from across the internet for free to train their models, but do not allow others to learn from their models. Nadella considers this unfair. He believes that if labs use open data, enterprises should also have the right to create their own smaller and cheaper models based on the results obtained from these models (distillation).

Currently, companies like Anthropic are complaining that their model outputs are being harvested by other competitors, particularly Chinese open-source models. However, Nadella called the model creators' restrictions on this matter "ironic." He believes that tech giants must also adhere to principles of openness.

In conclusion, Satya Nadella's statement signals that a new era of regulation is beginning in the AI market. Companies now need to think seriously not only about how to use AI capabilities but also about how to protect their most valuable asset—their data. Otherwise, today's assistant could easily become tomorrow's strongest competitor.