ai risk today logo svg bl font
Menu
SUBSCRIBE
SUBSCRIBE
ai risk today logo svg bl font
Menu

Generative AI

Visualization of AI governance and browser activity illustrating violations of company AI policies.

48.3% Of U.S. Employees Admit To Breaking Company AI Policies

, , , , , , , , , , , ,

Nearly seven in 10 employees also reported pasting sensitive company data into AI tools, suggesting many organizations face an enforcement problem rather than an awareness problem.

Key Takeaways

  • 48.3% of U.S. employees knowingly breach their company’s AI policy.
  • 68.7% said they pasted internal, sensitive, or regulated data into AI tools in the last three months.
  • 70.9% use AI tools daily for work.
  • 42.3% reported using approved and unapproved AI tools side by side during the same browser session.
  • 49.3% said they would use an unapproved AI tool if their preferred tool was not authorized.

Neon Cyber’s report, Quantifying Shadow AI Risk in the Browser, suggests many organizations face a growing AI governance challenge that extends beyond policy development. The survey of 227 U.S. employees who use AI at work found that nearly half knowingly violate their employer’s AI policies despite understanding those rules.

The report argues that the problem is no longer whether organizations have AI policies. Instead, it suggests many organizations struggle to enforce them where employees actually use AI.

Policy is not changing behavior

Among respondents, 63.0% said their organization has a clear AI policy they understand. Of those employees, 48.3% admitted they knowingly breached that policy.

The report also found that 39.6% knowingly used unapproved AI tools, while another 15.0% said they might have. Nearly half (49.3%) said they would simply switch to another unauthorized AI tool if the one they wanted was not approved by their employer.

According to the report, employees often prioritize productivity over internal approval processes, particularly as AI tools become a routine part of daily work.

Sensitive company data continues flowing into AI

The survey found that 68.7% of respondents pasted or uploaded internal, sensitive, or regulated information into AI tools during the last three months, including financial information, customer data, and login credentials.

Shadow AI is becoming routine

More than 70% said they use AI tools daily, while 51.1% reported using them multiple times each day. Nearly two-thirds described AI as essential or very necessary to their jobs.

At the same time, 42.3% said they use both approved enterprise AI tools and unapproved AI services in the same browser session, using the same corporate data. The report argues that this makes it difficult for organizations to detect unauthorized AI activity using traditional security controls.

Browser habits also contributed to the risk. Nearly 70% reported using the same browser profile for both work and personal activities, while 63.0% said they mix personal and work accounts within the same browsing session.

Neon Cyber conducted the survey in May 2026 among 227 self-identified U.S. employees who use AI at work.

IBM buys Confluent for better enterprise AI capabilities

IBM to Acquire Confluent In $11B Deal To Power Enterprise Generative AI

, , ,

The tech giant will pay $31 per share to acquire Confluent’s real-time data streaming infrastructure and integrate it into its cloud and AI services.

 

IBM and Confluent announced they signed a definitive agreement under which IBM will acquire Confluent in an all-cash transaction valued at approximately $11 billion. Under the terms of the deal, IBM will pay $31 per share for all outstanding Confluent stock.

Confluent provides data streaming software based on the open-source Apache Kafka project, which is used to process and move data in real time across cloud and on-premises systems. The company reported serving more than 6,500 customers worldwide, including large enterprises, through its managed cloud platform and related software offerings.

IBM said it will fund the acquisition using cash on hand. The transaction has been approved by the boards of both companies, and shareholders representing approximately 62% of Confluent’s voting power have agreed to support the deal. The acquisition is expected to close by mid-2026, subject to regulatory approvals and other customary closing conditions.

IBM Chief Executive Arvind Krishna said the acquisition is intended to strengthen IBM’s data and AI capabilities, particularly as customers deploy generative and agent-based AI systems that require access to real-time data. IBM said it would integrate Confluent’s technology into its existing cloud and AI infrastructure.

The deal continues IBM’s recent focus on expanding its software portfolio, particularly in cloud and automation tools. It follows IBM’s acquisition of HashiCorp last year and reflects a broader effort to position the company around hybrid cloud and AI-driven services.

Confluent shares rose sharply following the announcement, gaining more than 20% in after-hours trading, while IBM shares were little changed. Analysts noted that the transaction gives IBM greater exposure to enterprise data infrastructure, while raising questions about how the two companies’ platforms will be integrated and how regulators will review the acquisition.