Threat Labs Report: Manufacturing 2025

GenAI adoption in the manufacturing sector has remained consistently high over the past year, with usage fluctuating between 90% and 97%. Currently, 94% of organizations use genAI applications directly, 97% use apps that leverage user data for model training, and 96% use genAI-powered tools indirectly. This reflects a stable and mature adoption trend across the sector, aligning closely with global usage patterns.

At the same time, the use of personal genAI accounts remained relatively stable at around 83% through December 2024, before gradually declining to 51% by September 2025. In contrast, adoption of organization-approved genAI solutions rose notably over the same period, increasing from 15% to 42%. This shift highlights a growing preference within the manufacturing sector for company-approved platforms that offer stronger safeguards for sensitive data and tighter governance over genAI usage. As this transition continues, organizations should focus on adopting secure, enterprise-grade genAI solutions that foster innovation while maintaining compliance and minimizing risk.

In the manufacturing sector, the top 10 most widely used genAI applications largely mirror global adoption patterns. ChatGPT leads with 87% of organizations using it, followed by Google Gemini at 74%, showing strong adoption across the sector. Microsoft 365 Copilot is used by 58% of organizations, while Microsoft Copilot follows closely at 56%, reflecting growing demand for genAI solutions embedded within productivity and workflow platforms. The remaining top applications include a diverse mix of domain-specific and embedded AI tools designed to address specialized operational and industrial needs.

The chart below illustrates how the popularity of the top 10 genAI applications has evolved over the past year, underscoring the continued transformation of the genAI landscape within the manufacturing sector. ChatGPT remains the most widely used application, maintaining a steady average usage of around 85% throughout the year. Google Gemini has shown notable growth, increasing from 51% to 75%, signaling rising adoption and interest. Microsoft 365 Copilot has also continued its upward trajectory, now reaching 58%, driven by its integration into core productivity and enterprise tools. Grok has emerged as a strong new entrant, rising from near-zero adoption in March to 32% by September, reflecting growing experimentation with alternative genAI platforms in the sector.

As genAI adoption continues to expand across the manufacturing sector, concerns around data exposure are becoming increasingly critical. Organizations are using genAI tools for a range of functions, including summarizing technical documents, generating reports, and optimizing software development workflows. However, these use cases inherently involve sharing potentially sensitive data with genAI applications, expanding the attack surface for data security threats.

Data protection has therefore become a key priority, especially as genAI tools become deeply embedded in daily operations and shadow IT remains a persistent challenge.

Recent analyses of data policy violations show that the most frequently exposed category of sensitive information in the manufacturing sector is regulated data, accounting for 29% of all incidents. Source code follows closely at 28%, often shared inadvertently when developers use genAI to write, review, or summarize code. Passwords and API keys represent 26% of exposures, reflecting ongoing risks tied to insecure prompt usage and lack of data governance. Overall, these exposure patterns align closely with global trends, underscoring the need for stronger DLP controls and secure deployment of enterprise-grade genAI tools.

Organizations across the manufacturing sector are adopting a cautious stance toward genAI tools, with many choosing to block specific applications over security, privacy, and compliance concerns. While policies differ across companies, certain tools are restricted far more frequently than others—reflecting where organizations perceive the highest risk. For those managing similar environments, it may be worth considering whether blocking entire categories of apps offers stronger protection than managing each one individually.

Within the sector, DeepSeek ranks as the most frequently blocked genAI application, with 48% of organizations restricting access due to concerns about transparency and the risks associated with rapidly emerging genAI platforms. ZeroGPT is at 43%, largely because of its data handling practices, including storage of submitted content and reported redirections of information to third-party sites.

These blocking patterns indicate that organizations in the manufacturing sector are not only responding to risks posed by individual applications, but also strengthening their overall governance strategies for managing genAI within established security and compliance frameworks.