When cybersecurity compliance increases exposure

Beijing has rebranded routine cybersecurity practice into something the state can direct and supervise. When the enforcement of regulatory rules is discretionary and framed as an issue of national security, multinational organisations operating in China who improve their cybersecurity beyond baseline compliance may enable the state to gain insight into operations, data, and the behaviour of individuals within the company. This  greater visibility can provide valuable intelligence and bargaining leverage for the state, rather than reduced cyber risk.

Frameworks such as China’s Multi-level Protection Scheme (MLPS 2.0) tie system classification to security obligations and enable the inspection of corporate systems. MLPS 2.0 classifies information systems into five levels, each potentially giving the state more access to technical solutions, data, and personal details collected by the company in efforts to secure exactly that information. The regulation is unique in being mandatory for any company operating a network inside China and requires enterprises to undergo self-assessments, state-appointed expert verifications, and filing with local authorities. 

Separately, ‘important data’ must undergo security assessment before export under China’s data laws. Because both inspection thresholds and the definition of ‘important data’ retain a high degree of state discretion, the risk for firms lies less in any single regulation than in the uncertainty of how ordinary security activities may later be reclassified and acted upon by the state. 

What matters is not inspection alone, but inspection combined with uncertain thresholds, evolving data categories, and broad state authority. At stake is also data sovereignty: the state’s ability to determine where operational data resides, how it is classified, and when access to it shifts from a matter of corporate governance to one of national security.

The monitoring paradox

This tension becomes most visible in security monitoring. Modern cybersecurity depends on centralised visibility – aggregating logs and signals, and detecting anomalies across an enterprise. Without it, firms cannot effectively respond to cybersecurity incidents. China’s MLPS 2.0 mandates a framework based on the concept of ‘one center and three lines of defense’, which arguably improves an organisation’s cybersecurity, but also amplifies the state’s visibility into said organisation. Instead of allowing logs to remain siloed, for organisations of level 3 or higher, the framework mandates the consolidation of logs of user behaviour, system events, and traffic flows into a central security management layer. 

When paired with the Cybersecurity Law of China’s (CSL) requirements to retain time-synchronised network logs for at least six months, regulatory authorities gain access to a highly standardised, chronological, easily auditable blueprint of an organisation’s digital operations.

The paradox is that the same measures that improve security can also potentially give a foreign government insight into an organisation’s intellectual property and user behaviour. Often, these measures aim to protect said data from state-sponsored actors or APTs, but, through compliance, they render a complex, proprietary corporate network orderly, standardised, and therefore easily mapped from the outside. 

This compliance mechanism introduces a major vulnerability to corporate assets. Certified third-party institutions subject systems and products to periodic security testing and certification, which may include deep technical review, potentially exposing data in sensitive sectors and putting non-Chinese intellectual property at risk. Centralised monitoring can also trigger cross-border restrictions when logs include regulated data. 

These regulatory measures do not only increase visibility, but also create leverage by enabling the state to enforce compliance, selectively restrict access, and impose operational or economic costs. In practice this leverage has been exercised through licensing regimes, real time audit access, data localisation requirements, and the ability to suspend or degrade services that fail to meet dynamic regulatory thresholds. 

Prevailing assumptions, especially the idea that better technical controls inherently reduce risk, require scrutiny. Where best-in-class security expands supervisory access, optimisation may alter or increase exposure rather than eliminate it.

AI in cybersecurity management erodes boundaries

The risk extends into the use of non-Chinese AI in security monitoring practices in China, which would trigger compliance events because of the ingest, processing, and ‘export’ of sensitive data of Chinese nationals to the mandated central security layer. 

Effective security monitoring relies on the auditability and traceability of actions, linking the who to the what, when, and where. If logs, identities, or behavioural information that could qualify as personal or ‘important’ data is centralised or transferred abroad, firms may trigger government review, security assessment, or regulatory inspection increasing their exposure. 

Similarly, these dynamics mean that using AI for cybersecurity under CSL and MLPS 2.0 expands what must be observable, standardises how it is monitored, and deepens regulatory access into user and system behaviour, thereby converting technical visibility into institutional leverage. 

Additionally, generative AI (GenAI), which is increasingly embedded into enterprise tools, accelerates the exposure of sensitive data beyond what firms can control. By 2027, 40% of AI-related breaches could result from cross-border GenAI misuse. Prompts, outputs, and embedded features in software-as-a-service tools can route data through opaque vendor chains and shifting processing locations. Under China’s evolving rules, downstream processing can be treated as an unauthorised cross-border transfer and result in state inspections, warnings, or fines.

Implications and options

The core challenge is structural. China’s cybersecurity and AI frameworks allow for state discretion over which data is deemed sensitive, what constitutes an incident, and when oversight escalates from compliance review to operational supervision.

For multinational firms, the implication is not to abandon strong security practices, but to rethink their objectives. Separation between China-based systems and global infrastructure becomes necessary, even if it reduces efficiency. Limiting what data is collected and retained can also reduce exposure. This could include an independent, Chinese active directory, localised cloud deployments, and dedicated local security infrastructure so as to comply with the mandated ‘one center and three lines of defense’ framework. These are not ideal technical solutions, but they reflect the constraints of the environment.

China’s approach does not rely on constant intervention. Its leverage lies in uncertainty and in the ability to redefine thresholds and expand its oversight when conditions warrant. For multinational firms, the task is not perfect compliance. It is resilience in a system where compliance can become control, and technical decisions carry strategic consequences.