Data security 2025 shows breach exposure outpacing control maturity

At a glance

What this is: This retrospective argues that 2025 showed breach exposure, AI leakage and supply chain weakness moving faster than most organisations' data controls.

Why it matters: For IAM, NHI and broader security teams, the lesson is that visibility, entitlement control and third-party governance now determine whether data protection survives real-world pressure.

Context

2025 exposed a governance gap in data security: organisations can accumulate privacy obligations faster than they can operationalise control over data, AI usage and third-party access. The article frames this as a shift from policy compliance to continuous security management, where visibility and containment matter more than periodic review.

The article is especially relevant to identity and access governance because many of the failures described involve who or what can reach sensitive data, whether that is an employee using AI tools, a compromised supplier, or a malicious package introduced into a build chain. In that sense, the data problem is also an access problem.

The pattern is not atypical for a year of rapid AI adoption and persistent supply chain abuse: when data is highly distributed, the control plane tends to lag the business plane.

Key questions

Q: What breaks when employees use unapproved AI tools with company data?

A: Governance breaks because the organisation loses visibility into where data and secrets are going, who can access them, and how they are being reused. Unapproved tools can copy credentials into unmanaged workflows, which weakens revocation and makes audit trails incomplete. The result is shadow access outside the main identity programme.

Q: Why do third-party services and packages increase data exposure risk so quickly?

A: They extend trust outside the organisation's own control plane, often using broad permissions, embedded secrets or unreviewed update paths. When those dependencies are compromised or misused, they can expose credentials, personal data and sensitive files across many downstream systems at once. The risk is amplified when access is long-lived and poorly inventoried.

Q: What are the signs that data security controls are failing across an organisation?

A: Common warning signs include unexplained data access spikes, sensitive files moving outside approved channels, poor audit trail coverage, repeated policy exceptions, and inconsistent access reviews. Teams should also watch for delayed detection of phishing, over-privileged accounts, and gaps between data classification and actual controls, because those gaps usually indicate the programme is not keeping pace with risk.

Q: Should organisations prioritise visibility or containment first for data security?

A: They need both, but containment should follow immediately when visibility reveals sensitive data in risky places. Visibility without containment only documents exposure, while containment without visibility misses the next source of leakage. The practical order is discover, classify, constrain and then verify that the same data cannot reappear in uncontrolled paths.

Technical breakdown

Why breach exposure keeps outpacing data control maturity

Data control maturity is the ability to discover, classify, govern and contain sensitive information across systems, users and services. This article shows the mismatch between that maturity and the speed of modern exposure, where breaches, AI usage and third-party dependencies all expand the attack surface faster than policy updates can catch up. The issue is not merely volume of incidents, but the failure of point-in-time controls to keep pace with dynamic data movement across internal and external ecosystems.

Practical implication: shift from periodic data reviews to continuous discovery, monitoring and containment across all high-risk data paths.

How AI becomes a data leakage channel

AI tools create a distinct leakage path because users can paste sensitive material into systems that may process, store or reuse that content outside intended boundaries. The article notes that many organisations experienced AI-related leakage and that model or application breaches were often tied to weak access controls. In practice, this is an identity and governance problem as much as a model-risk problem, because the control failure begins at access approval, usage policy and data egress rather than only at inference time.

Practical implication: govern employee AI use with access policy, data classification and logging controls, not awareness messaging alone.

Supply chain exposure now includes packages, platforms and file-transfer services

Supply chain compromise in data security is no longer limited to upstream vendor outages. The article highlights unpatched third-party services and malicious packages in repositories such as npm and PyPI, both of which can become channels for data theft and credential exposure. That broadens the control problem from vendor due diligence to runtime trust, dependency hygiene and secret containment, especially where build and transfer tooling handles sensitive information by default.

Practical implication: treat third-party access paths, package intake and file-transfer tooling as high-risk data corridors requiring tighter entitlement and secret controls.

Threat narrative

Attacker objective: The attacker seeks to extract sensitive data at scale and use that exposure for extortion, resale, downstream compromise or operational disruption.

1. Entry occurs through compromised third-party software, malicious packages or unsafe employee use of AI tools that interact with sensitive data.
2. Credential and secret exposure follow when those channels reveal access material or permit broader retrieval of protected information.
3. Escalation happens as attackers use the exposed data, credentials or trusted integrations to reach additional systems, suppliers or user records.
4. Impact includes mass data exposure, regulatory exposure, extortion leverage and business disruption across internal operations and supply chains.

Breaches seen in the wild

• Shai Hulud npm malware campaign: Shai Hulud campaign: npm malware exposed secrets on GitHub.
• reviewdog Action compromise 2025: A stolen maintainer token poisoned reviewdog/action-setup, leaking CI secrets including the tj-actions bot token used in the next attack.

Read and download The State of NHI & AI Agent Breach Report 2026, covering 200+ breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Data security debt is now a governance problem, not just a tooling problem. The article shows that organisations are collecting legal obligations, AI usage risks and supplier exposure faster than they are closing the control gap. That creates a form of data security debt, where policy intent and operational reality diverge until the breach forces reconciliation. Practitioners should treat visibility and containment as standing governance requirements, not project outcomes.

AI has become a direct leakage channel for sensitive data. The article's figures on employee AI use and AI-related breaches point to a pattern where the access decision is the real control point. If employees can move sensitive content into unmanaged AI tools, the issue is not model intelligence but the absence of approved data pathways, logging and usage constraints. Security teams should recognise that AI governance now overlaps with identity governance at the point of use.

Supply chain visibility is the new perimeter for data protection. Third-party services, package repositories and managed file transfer platforms all act as trust accelerators, but also as leakage amplifiers when controls are weak. This is especially relevant to NHI governance because packages, services and integrations often authenticate with long-lived secrets and broad permissions. Practitioner conclusion: inventory and constrain every data-bearing trust path before it becomes a breach path.

Privacy regulation has outgrown point-in-time compliance models. The article cites 155 countries with data protection and privacy legislation, yet the operational breaches it describes still recur because governance is not embedded in day-to-day control. Compliance alone does not prevent data exfiltration, model misuse or third-party leakage. The field now needs continuous enforcement and evidence-based control assurance, not annual certification cycles.

Data security maturity will increasingly be measured by containment speed. The organisations that reduce loss will be the ones that can discover sensitive data quickly, limit exposure windows and prove who or what touched it. That matters for human access, NHI credentials and AI-driven workflows alike, because the same governance failure repeats across all three. Practitioners should optimise for rapid containment, not just after-the-fact visibility.

From our research library:

• 92% of organisations expose NHIs to third parties, raising concerns about supply chain security, according to the Ultimate Guide to NHIs.
• Breaches involving third parties rose to 48% of all breaches, a 60% increase on the previous year, according to Verizon's 2026 Data Breach Investigations Report.
• Read next: AI Supply Chain Security and AI-BOM Guide

What this signals

Data security debt: the gap between legal coverage and operational control is widening, and that gap will keep showing up in breach patterns until discovery and containment become continuous rather than episodic. Practitioners should assume that policy proliferation does not equal data protection maturity.

The strongest signal for 2026 is that AI, suppliers and software dependencies now function as a single exposure surface. Organisations that cannot trace where sensitive data goes, who can reach it and which tools can exfiltrate it will keep discovering the same failure in different forms.

The article's 155-country privacy backdrop shows the regulatory bar is rising, but the more immediate operational question is whether security teams can prove containment under real-world usage, especially where AI and third-party access intersect.

For practitioners

• Implement continuous sensitive-data discovery Map where regulated and business-critical data lives, how it moves and which services can access it. Use continuous discovery rather than periodic audits so new stores, replicas and shadow workflows are surfaced before exposure becomes systemic.
• Lock down employee AI data pathways Create approved AI usage patterns for staff, define which data classes can never be pasted into external tools and log high-risk interactions. Pair policy with technical enforcement so the control applies where data is actually entered.
• Treat third-party data paths as privileged Review supplier integrations, file-transfer services and package dependencies as trust channels rather than generic tooling. Reduce standing access, restrict secrets exposure and validate that external components cannot reach more data than their function requires.
• Contain secrets in build and repository workflows Scan repositories, dependency feeds and CI pipelines for credentials, tokens and sensitive file contents. Remove hardcoded secrets, rotate any exposed values and require short-lived credentials for systems that handle sensitive data during software delivery.

Key takeaways

• 2025 showed that data breaches, AI leakage and supply chain compromise are converging faster than many organisations can update their controls.
• The article's own figures point to a structural gap between regulatory expansion and operational visibility, especially where AI tools and third-party services handle sensitive data.
• The practical response is continuous discovery, tighter containment and stronger governance over who and what can move sensitive information.

Key terms

• Identity Security Debt: The accumulation of unresolved identity control gaps across authentication, authorisation, lifecycle, and exception handling. In finance, it shows up when old access patterns remain in place while new controls are layered on top, leaving the programme looking modern but still carrying inherited risk.
• Sensitive Data Discovery: Sensitive data discovery is the process of locating where protected or regulated information exists across systems, storage, and workflows. In cloud environments, it must be continuous because assets appear, move, and replicate quickly, making one-off inventories unreliable for governance or incident response.
• AI data leakage: AI data leakage occurs when sensitive business information is exposed through prompts, outputs, or copied content in AI-assisted workflows. In browser-driven work, the risk is often accidental rather than malicious, so governance depends on data rules, usage policy, and session controls.
• Third-party data path: A third-party data path is any route through which external vendors, services or software dependencies can process or transmit organisational data. These paths matter because they often carry inherited trust, making visibility, permission scope and offboarding controls central to security.

Deepen your knowledge

The NHI Foundation Level course, the industry's only accredited NHI security programme, covers NHI governance, machine identity security and secrets management for teams managing modern access risk. It is designed for practitioners who need to connect identity control to broader security operations and governance.