The increasing use of artificial intelligence (AI) by large companies is raising concerns about how data is managed when using external AI models. As businesses rely more on generative AI—systems that create text, images, or other content—ensuring the security of internal information, customer data, and trade secrets has become a top priority. Companies like Palantir and NVIDIA have responded by tightening controls on how they use AI models from external providers such as OpenAI and Anthropic. Palantir, for instance, has asked Anthropic to implement a "Zero Data Retention" policy, which would prevent the storage of any user data. Palantir is even considering temporarily blocking access to certain Anthropic models until these conditions are met. NVIDIA has taken a different approach by limiting the use of Anthropic models to tasks that involve minimal sensitivity. For internal operations, the company is relying on its own AI models, such as Nemotron. This concern is especially acute for firms working with government or defense agencies, where preventing the transfer of confidential data to external AI services is crucial. In these sectors, using generative AI can involve sending data to infrastructure that is not directly controlled by the company. As a result, many organizations are favoring solutions that offer more control, such as isolated systems, on-site installations, or AI models that run directly on the company’s own infrastructure. Anthropic has updated its policies regarding the retention of user data, allowing for a maximum of 30 days of data storage. This measure is intended to help detect and prevent advanced cyber threats. Both OpenAI and Anthropic emphasize that, by default, data from professional clients is not used to train their models unless explicit permission is given. However, some technical data or anonymized metadata may be processed to improve their services, depending on the terms of use. OpenAI has faced scrutiny over how its models are trained, particularly after comments suggesting that AI could be used to solve complex scientific challenges. These discussions have intensified the ongoing debate about the transparency of data used in AI systems. Microsoft is also addressing these concerns by developing cloud environments tailored to the security and governance needs of businesses. In addition, the company is promoting its own AI services to meet the growing demand for secure, customizable AI solutions. As companies continue to balance the benefits of AI with the risks of data exposure, the push for greater control over AI infrastructure is likely to shape the future of enterprise AI adoption.