A new report highlights how AI tools are becoming a central part of everyday work, especially in security. Companies are moving beyond testing AI for fun or simple chats. Instead, AI is now embedded into workflows, connected with core business tools, and capable of acting on its own. This rapid shift raises important questions about how organizations keep their data safe and maintain control over AI systems.

Widespread Use of Leading AI Models

Most organizations now use major language models like OpenAI, which is present in 96% of companies, and Anthropic, seen in nearly 78%. These AI providers form the backbone of many AI activities in the workplace. They are integrated with various tools and systems, making AI a common part of daily operations.

Beyond chat, AI is expanding into different areas. Meeting tools like Otter.ai are used by over 74% of companies for transcribing conversations. Presentation tools such as Gamma are used by more than half of organizations, while coding assistants like Cursor and voice tools like ElevenLabs are also gaining popularity. This diversification shows AI’s growing role across multiple functions.

Emerging AI Agents and Integration Trends

New types of AI tools, called agentic tools, are starting to appear. These are designed to act autonomously or semi-autonomously to perform tasks. Examples include Manus, Lindy, and Agent.ai, which are still early in their adoption but show potential for more advanced automation.

Organizations are integrating AI into their existing systems frequently. OpenAI and Anthropic’s models are commonly connected to productivity suites, knowledge management systems, and code repositories. These integrations make it easier for AI to assist in various workflows, but they also introduce new security considerations.

Usage patterns show that OpenAI’s models account for about 67% of prompt activity, indicating their dominance. However, prompts often involve copying data or uploading files, which can pose risks. About 17% of prompts include copying or uploading sensitive information, creating potential exposure points.

Risks and Challenges in AI Security

Data leaks are a real concern. The report found that many security events involve sensitive data like secrets, credentials, financial details, and health information. Secrets and credentials make up nearly half of these incidents, highlighting the need for better data controls.

This research was based on anonymized data from Nudge Security’s customers, not surveys. The findings reflect actual AI activity within enterprise environments, providing a clear picture of current usage and risks. It shows how AI adoption is progressing and where organizations need to focus their security efforts.

Despite the growing importance of AI governance, many security programs still focus mainly on approving vendors or setting policies. But as AI becomes more autonomous and embedded into critical systems, organizations must shift to continuous, real-time monitoring. Effective governance now requires an adaptive approach that keeps pace with AI’s rapid evolution.

Inspired by

• https://ai-techpark.com/nudge-security-study-ai-adoption-reshapes-security-governance/