Understanding the Rise of AI Agents in Multi-Platform Environments: Insights from Nudge Security Research

The Multi-Platform Rise of AI Agents: Insights from Nudge Security



In a groundbreaking report, Nudge Security, a pioneer in SaaS and AI security governance, has unveiled significant findings regarding the adoption and functionality of AI agents within organizational settings. Released on November 17, 2026, the report titled "AI Adoption in Practice 2026 Beyond the Approved List" provides a comprehensive view of how AI integration is reshaping workplace technology use. This research is based on anonymized telemetry data collected from Nudge Security's customer environments, ensuring reliability beyond traditional surveys.

Key Findings


One of the most striking statistics from the report is that more than half of the organizations currently employing AI agents operate them across two or more platforms. Key players such as Atlassian, Claude, Cursor, OpenAI, and Zapier are at the forefront, highlighting the extensive reach of AI in daily business functions. Nudge Security co-founder and CTO Jaime Blasco points out that business platforms are organically evolving into agent platforms, regardless of formal approval or oversight. This development introduces critical challenges for security teams, necessitating a shift in focus towards understanding what tools are operational rather than merely which are approved.

The Growing Use of AI Agents


The report reveals that a significant portion of AI agents functions under employees' identities; specifically, half of the organizations using AI deploy seven or more agents, with over 40% managing more than ten. These agents often act within contexts like project management systems and CRM tools. The immediate implications are clear: as these AI tools are interwoven with employee access to company data, the opening for potential security vulnerabilities widens.

Direct Connections to Sensitive Systems


Moreover, the findings show that AI tools are connecting directly to highly sensitive organizational systems. Alarmingly, 74% of organizations have integrated with at least one AI platform from Anthropic, and 67% are engaged with OpenAI. These integrations facilitate direct access to email, code, chat, and documents, bypassing traditional inspection mechanisms that could flag unauthorized data transfers.

AI Accessibility and Market Dynamics


The report highlights a significant trend: treating AI models as commodities. Over 91% of surveyed organizations utilize both OpenAI and Anthropic, with Anthropic's adoption seeing a notable increase. Additional insights reveal that a growing number of AI tools—twelve in total—are found in more than half of the surveyed organizations, a stark rise from just five reported earlier in January this year. The dynamic nature of AI integration has led to the emergence of more than 27 distinct tools across participating environments.

Employee Interactions with AI


Around 67% of organizations with at least one AI browser extension reported utilizing tools like Grammarly and Claude, both of which can access content from employee-surfed web pages. This level of integration raises questions about data privacy and security, especially as one in five observed prompts include sensitive data being uploaded or pasted into AI interfaces. Although OpenAI remains dominant, its share of prompts has decreased, indicating a diversification toward other AI tools.

Recommendations for Security Leaders


Given the revelations in this report, the authors urge security teams to rethink their strategies in managing AI risks. Visibility alone is insufficient; the report offers several actionable steps:
1. Inventory Agents Thoroughly: Security teams should track the agents built within the platforms their teams utilize, identifying who created each agent and whose credentials they operate under.
2. Evaluate OAuth Grants: Treat OAuth access permissions as significant risk factors by assessing which AI tools hold access to sensitive information and revoking unnecessary permissions.
3. Shift Focus from Approved Lists: Move from merely observing which tools are approved to understanding what they connect to, thus providing a more nuanced view of associated risks.
4. Scrutinize AI Browser Extensions: Review extensions capable of reading page content and ensure clear guidelines for their use in sensitive environments.
5. Engage the Workforce: Recognize that employees often understand their AI integrations well; involve them in dialogue to better navigate risks associated with AI tools.

By incorporating these recommendations, organizations can navigate the complexity of growing AI adoption while safeguarding company data and maintaining productivity. The full insights of this report serve as a call to action for security leaders to be proactive in their strategies for managing the evolving landscape of AI and workplace technology.

Conclusion


Nudge Security’s findings illuminate the pressing challenges organizations face as AI agents become increasingly prevalent across numerous platforms. As these technologies evolve, understanding and managing associated risks will be crucial for maintaining security in workplaces that thrive on innovation and efficiency.

Topics Business Technology)

【About Using Articles】

You can freely use the title and article content by linking to the page where the article is posted.
※ Images cannot be used.

【About Links】

Links are free to use.