Shadow AI is creating an unmanaged identity crisis

Shadow AI is creating an unmanaged identity crisis
Image Credit: Keeper Security

Across APAC, organisations are unwittingly populating their work environments with unmanaged and unmonitored non-human identities.

By on

For over a decade, enterprise security teams believed they understood the rules of shadow IT. An employee downloaded an unauthorised file-sharing tool or spun up an unsanctioned SaaS dashboard, and IT responded with web gateways, firewalls, and application blacklists.

But shadow AI breaks that familiar playbook entirely. Unlike static software of the past, modern AI platforms are deeply integrated, agentic workflows. When teams link AI assistants to enterprise calendars, cloud repositories and CRM platforms to accelerate everyday work, they are not just using an unapproved application – they are granting persistent credentials, API keys and elevated data access to machine actors.

Some of the consequences are already measurable: IBM's 2026 Cost of a Data Breach Report found that 43 percent of breached organisations had experienced an incident involving shadow AI, up from 20 percent a year earlier, and those breaches involving shadow AI cost an average of US$5.39 million.

The conversation around shadow AI has spent too long focusing purely on data exposure when the far more systemic crisis is that organisations are quietly populating their environments with unmanaged and unmonitored Non-Human Identities (NHIs).

Some recent data exposures across the APAC region highlight these operational risks. For example, Panasonic Holdings in Japan disclosed that an employee uploaded personal records of 334,000 persons to an unsanctioned generative AI tool. Food manufacturer and retailer Bee Cheng Hiang in Singapore saw over 95,000 customer email addresses exposed by an unreviewed AI-generated bulk mail script. And over in Australia, a contractor uploaded a spreadsheet to ChatGPT, exposing sensitive personal and health records of 3,000 flood recovery applicants to the NSW Reconstruction Authority.

Each AI integration creates a new identity

Modern AI tools require little or no setup, and employees can connect applications, automate workflows and process data within minutes. That ease of adoption allows shadow AI to spread at a pace few governance frameworks were designed to match.

Once an AI platform gains access to third-party applications, cloud environments and internal databases, it introduces credentials and identities that sit beyond the view of the teams responsible for protecting them.

Every account an employee creates for an AI tool is an identity with its own access scope, data permissions and session history. Connecting those tools to internal systems introduces service accounts that typically sit outside traditional lifecycle management, credential rotation and access governance.

Each integration also generates API keys, tokens and secrets, which often end up in browser extensions or configuration files beyond the security perimeter. The resulting NHIs accumulate across external platforms without a clear owner and remain active long after the employee has left the organisation.

The governance gap across APAC

Traditional identity security assumes human users, IT-provisioned access and defined network perimeters, but shadow AI undermines all of those assumptions at once.


 An AI agent with access to a production database carries the same privileged access risk as a human administrator with equivalent permissions – and it operates at machine speed.

-Takanori Nishiyama, Senior Vice President, APAC & Japan Country Manager, Keeper Security 

Organisations should apply the same oversight they use for privileged administrators, including session review, access approval and regular recertification.

Our recent Identity Security at Machine Speed report surveyed 3,200 cybersecurity decision-makers and senior IT leaders, including 500 in APAC across Australia, India, New Zealand and China. In the region, 53 percent of respondents identify AI-related NHI management as a top gap in their identity governance, 10 points above the global average.

Shadow AI compounds the problem, as 47 percent of APAC respondents cite a lack of visibility into the AI tools employees use as a significant governance gap. Just 38 percent of APAC organisations have fully deployed a Privileged Access Management (PAM) solution.

APAC regulators are setting the frameworks and guidelines

Policymakers across APAC increasingly treat AI agents as identities that need identities, owners and limits. In Japan, the Ministry of Economy, Trade and Industry (METI) and the Ministry of Internal Affairs and Communications (MIC) published version 1.2 of the AI Guidelines for Business in March this year, an update that introduces a definition of AI agents.

Singapore's Infocomm Media Development Authority (IMDA) launched the Model AI Governance Framework for Agentic AI earlier in January, calling for each agent to carry a traceable identity linked to an accountable human, and recommending that organisations restrict agents' tools and data to the minimum necessary.

The Cyber Security Agency of Singapore (CSA) followed in June with its finalised Addendum on Securing Agentic AI. In Australia, the National AI Centre's Guidance for AI Adoption asks organisations to decide who is accountable and to maintain human control, while the Office of the Australian Information Commissioner (OAIC) recommends that organisations avoid entering personal information into publicly available generative AI tools.

These frameworks remain voluntary, yet they set the expectations that boards and auditors will apply. Organisations that align their identity controls with them now will find compliance far easier than those that wait for binding rules.

What organisations should do now

Blocking AI tools outright will not solve shadow AI. Employees who lose access to a sanctioned tool will route around the block, and the identities they create will be harder to see. The response requires organisations to extend identity-first security controls to every identity that AI tools create – both human and machine alike – and to make governance continuous rather than periodic.

Effective governance begins with visibility. Security teams should continuously discover every sanctioned and unsanctioned AI tool, agent and integration across their environments, and identify every NHI those tools create outside IT provisioning. Beyond discovery, they should monitor and record privileged sessions in real time, which allows them to audit AI-driven workflows as they happen instead of relying on retrospective reviews.

NHIs need the same discipline as human accounts. Every AI agent and service account should meet the same authentication, authorisation and lifecycle requirements as a human employee, with least-privilege access, automated credential rotation and deprovisioning that follows its human owner or workload. Where possible, organisations should move towards zero standing privilege by granting AI agents elevated access only for the task at hand and revoking it automatically afterwards.

Underpinning all of this is zero trust, which means verifying every access request, human or machine, and granting only the privileges each task requires. Enforced consistently, this approach turns an AI tool connected without approval into a visible, governed identity instead of a hidden entry point.

As AI becomes part of everyday work, shadow AI will continue to grow and drive a corresponding rise in unmanaged identities across the enterprise. Organisations need visibility into which AI tools reach their systems, and consistent governance over both human and machine identities alike.

With those controls in place, organisations can adopt AI at speed while retaining control over who and what is accessing their most critical systems.

Takanori Nishiyama is Senior Vice President, APAC & Japan Country Manager, Keeper Security

 

To reach the editorial team on your feedback, story ideas and pitches, contact them here.
© iTnews Asia
Tags:

Most Read Articles