Nearly half, 47%, of security decision makers1 have confirmed their organization has experienced an incident in which sensitive data was exposed through an AI tool, according to new research commissioned by BlackFog.

The findings show how speed is outpacing control, as organizations try to balance rapid AI deployment, with the need to protect sensitive data shared into AI tools. Amongst those organizations that had experienced a data incident more than three-quarters (77%) said a sanctioned, company-approved AI tool was involved.

The research also reveals that 39% of organizations report that AI adoption is moving faster than they can govern it. And, of those respondents feeling outpaced, only around a quarter (27%) have comprehensive AI controls in place.

The survey, conducted with UK and US cybersecurity leaders, also highlights the lack of visibility that organizations have into the types of data being shared with AI tools.

While the vast majority of organizations (98%) report having an AI usage policy in place, approaches to enforcement vary. Around a third (35%) have a written policy specifying what information employees must not enter into AI prompts, but without technical enforcement. Only just over half (56%) support their broader AI usage policies with comprehensive technical controls.

Additional key findings:

AI visibility falls short of what data is shared: Although 62% of respondents could identify users or departments sending data to AI tools, fewer than half of all respondents, 48%, could provide any record of the types of data that had left the organization through AI tools.

Autonomous AI agents present a gap in governance: More than half, 56%, of respondents say autonomous AI agents acting on behalf of employees fall outside their organization’s existing controls, suggesting governance may not yet be keeping pace with the shift from employees using AI tools, to AI acting on their behalf.

Within industry sectors this was highest among respondents in technology and telecoms with 70% of respondents stating that autonomous AI agents were outside of organizational controls.

Shadow AI is easier to spot than block: Only around half of respondents, 54% can block or restrict unsanctioned AI tools, whilst another 32% state that they can detect Shadow AI use but are unable to consistently stop it.

“Among organizations that reported sensitive data exposure, 77% said a company-approved AI tool was involved. This shows why security teams cannot stop at deciding which AI tools employees are allowed to use. Organizations also need to control what information is shared with them,” said Dr. Darren Williams, Founder and CEO of BlackFog.

“Written policies set expectations, but they cannot prevent sensitive data from leaving the organization on their own. Businesses need visibility into the data being shared and technical controls that can stop unauthorized disclosure at the point it happens. Those protections need to cover approved AI tools, unsanctioned services and autonomous agents acting on employees’ behalf.”

Controls are weakest around proprietary design and content

When asked about their ability to prevent risky behavior and the types of data shared, such as sales data, proprietary code, or IP, only around half of organizations can technically prevent it.

Controls are weakest around employees sharing proprietary design, creative or marketing content. For this type of content, 32% of respondents stated that they can detect or monitor, but can’t consistently prevent it, and a further 20% stated they rely primarily on written policies or training.

To learn more about the research, click here.

About BlackFog

BlackFog is the category-defining vendor in anti data exfiltration (ADX). Founded in 2015, the company invented ADX on the thesis that the endpoint is the only control point capable of stopping data from leaving an organization, an architectural bet that has now been validated across three exfiltration vectors: ransomware, shadow AI, and autonomous AI agents. BlackFog’s endpoint-native platform protects more than 500 enterprises, government agencies, and critical infrastructure operators worldwide. The company is the publisher of the annual State of Ransomware report and the BlackFog/Sapio Shadow AI Research, the most-cited primary research in the category. BlackFog’s recognition includes the teiss Awards 2026, the AI Excellence Award 2026, the Cybersecurity Excellence Awards 2026, and the Cybersecurity Breakthrough Award. Headquartered in San Francisco with international operations in London and Belfast. Learn more at blackfog.com

Notes for Editors

[1] Methodology

The results from this survey are from an online survey Sapio Research fielded on behalf of BlackFog with IT Security Decision Makers in companies of over 500 employees across the UK (200) and US (200). The research was conducted in September 2026.

Media gallery

About The Author
Business Wire