The Application of Artificial Intelligence in the Process of Intelligent

Criminal Justice: From the Perspective of Cross-Border Data

Governance

Xiqiao Tong

Law and Politics Department, North China Electric Power University, Baoding, Hebei, China

Keywords: Artificial Intelligence, Cross Border Data Governance, Criminal Justice, Algorithmic Ethics, International

Rules.

Abstract: In the context of global digital transformation, artificial intelligence technology is gradually being widely

applied by criminal justice institutions in various countries. While facilitating the law enforcement process of

domestic and international criminal justice, it also reconstructs the practical paradigm of transnational crime

investigation. However, its application in criminal investigation still faces certain institutional imperfections,

as well as numerous legal and ethical issues. This article aims to explore how artificial intelligence can be

correctly applied to the investigation of criminal cases from the perspective of cross-border data governance,

ensuring its legality, fairness, and effectiveness. Research has found that there is a certain lag in the current

WTO rules; Fragmentation of regional agreements increases the cost of judicial cooperation; Algorithm bias

poses a certain risk of misjudgement. Propose innovative paths such as establishing a data classification

system and creating an international AI investigation compliance committee.

1 INTRODUCTION

With the rapid development of artificial intelligence

technology, it is increasingly widely used in criminal

case investigation, especially in data analysis,

evidence collection, and suspect identification.

Artificial intelligence is changing the traditional

mode of criminal investigation. However, this

technological innovation has also brought new

challenges, and the application of artificial

intelligence in criminal investigation faces many

legal and ethical issues(Li,2020; Zhang,2021 ；

Joseph,2024).

At the same time, the rules for cross-border data

flow affect the sharing and cooperation of

transnational criminal data, and the limitations,

biases, and privacy protection principles of

algorithms pose higher requirements for the design

and application of artificial intelligence algorithms

(Dai,2023; Re, & Solow-Niederman,2019).

In addition, the principle of fair trade also requires

countries to maintain fair competition in the research

and application of artificial intelligence technology,

avoiding technological monopolies and unfair

competition.

This article adopts case analysis, comparative

research, and literature research methods to explore

how artificial intelligence can be correctly applied to

the investigation of criminal cases from the

perspective of cross-border data governance,

ensuring its legality, fairness, and effectiveness.

Firstly, by analyzing the main application scenarios

of artificial intelligence in criminal investigation, we

explore the efficiency improvement and potential

risks it brings; Secondly, study the regulatory blind

spots of laws in cross-border data flow, privacy

protection, algorithmic fairness, and the dynamic

balance of diverse value objectives; Finally, explore

how to promote the healthy development of artificial

intelligence in the field of criminal investigation

while ensuring security and privacy.

448

Tong, X.

The Application of Artiﬁcial Intelligence in the Process of Intelligent Criminal Justice: From the Perspective of Cross-Border Data Governance.

DOI: 10.5220/0014384300004859

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Politics, Law, and Social Science (ICPLSS 2025), pages 448-453

ISBN: 978-989-758-785-6

2 THE CURRENT APPLICATION

STATUS OF AI IN CRIMINAL

INVESTIGATION

2.1 Typical Application Scenarios

2.1.1 Cross Border Crime Warning

With the development of technology, police work has

gradually shifted from combating crime to focusing

on crime prevention. Therefore, determining which

type of person is a potential criminal before a crime

occurs is called crime warning (Blount, 2024). AI

integrates global data sources such as international

flights, hotel stays, and financial transactions,

combined with technologies such as facial

recognition and voiceprint comparison, to track the

biometric trajectory of suspects in real time. For

example, the Deep Seek platform covers a data

network of over 200 countries and can predict the

hiding preferences of fugitives, such as in Chinese

communities or remote areas. Machine learning

models can also analyze historical cases, construct

behavioral profiles of fugitives, and provide early

warning of potential criminal pathways.

Meanwhile, by utilizing blockchain analysis and

cross-border payment network monitoring, AI can

track the hidden asset paths of virtual currencies,

offshore companies, and other entities. For example,

using on-chain transaction graph analysis tools to

trace the money laundering behavior of privacy coins

such as Monero, and even crack anonymous identities

on the dark web.

In addition, AI can also perform sentiment

analysis and topic mining on social media content to

identify criminal threats. For example, extracting

geographical location clues from the dialect accent or

street view information in the background of short

videos to assist in locating suspects.

2.1.2 Collection of Electronic Evidence

AI can process large amounts of text, images, audio

and video data, extract key information, and classify

them (Barbir & Stankovic,2024). For example,

natural language processing (NLP) technology

analyzes chat records and email content, while image

recognition technology restores crime scenes or

identifies malware features. Multimodal models can

also mine hidden evidence from unstructured data,

such as encrypted files and deleted records.

At the same time, AI can construct personnel

relationship graphs, transaction networks, or time-

series logs through association analysis to discover

criminal patterns. For example, analyzing variables

such as customer education level and debt ratio in

loan fraud cases to identify risk points. Blockchain

technology ensures the integrity and immutability of

electronic evidence.

In addition, AI models such as Deep Seeker R1

can automatically generate case summaries, extract

elements such as involved personnel and fund flows,

and support parallel analysis of similar cases.

2.1.3 Judicial Cooperation

Intelligent systems (such as the artificial intelligence-

assisted trial system launched by Shenzhen Court)

unify the judgment scale, provide case

recommendations and sentencing recommendations,

and reduce the deviation of discretionary power. For

example, by analyzing key case data

comprehensively and generating judicial

recommendations to assist in social governance.

China and the "Belt and Road" countries have

combined satellite maps and other intelligent means

to solve the problem of monitoring blind areas. At the

same time, we will promote multilateral data security

cooperation mechanisms and promote the unification

of international judicial standards.

2.2 Focus on Legal Disputes

2.2.1 Data Sovereignty Conflict Issues

In cross-border criminal investigation, AI relies on

multi-source data such as biometric features, financial

transaction chains, and communication records,

which often involve data sovereignty disputes in

different jurisdictions. For example, the localization

requirement of the EU GDPR, which requires that

member states' data should not be transferred across

borders to countries that have not passed the

adequacy determination, has led to the need for

Chinese AI systems to establish localized data centers

in EU law enforcement cooperation, but may be

questioned for data sovereignty transfer. In addition,

the United States unilaterally requires companies to

provide overseas storage data through the Cloud Act,

which directly conflicts with the European Union's

Data Governance Act and forces cross-border

criminal investigation cooperation into a trade-off

between sovereignty first and technical efficiency.

There is a clear divergence in the positions of

China, the United States, and Europe regarding the

flow of cross-border data. China advocates data

sovereignty priority, that is, cross-border data flow

needs to be approved by the source country, and the

The Application of Artiﬁcial Intelligence in the Process of Intelligent Criminal Justice: From the Perspective of Cross-Border Data

Governance

449

localized AI governance template should be

promoted through the Belt and Road countries. The

EU, on the other hand, emphasizes compliance

barriers, which require AI system design stages to

embed data protection mechanisms (such as Article

35 of GDPR), leading non-EU technology suppliers

to restructure their algorithm architectures to comply

with European Standards. In addition, the United

States places greater emphasis on the technology

neutrality strategy, which involves restricting

technology exports to China through the Chip and

Science Act, while promoting the formation of a data

flow alliance among allies.

The conflict between Article 8 of the European

Convention on human rights (Privacy Issues) and AI

criminal investigation technology is prominent. AI

can invade cloud data through dark network cracking,

blockchain traceability and other technologies. The

traditional physical privacy boundary is invalid. The

European Court of human rights requires judicial

authorization in advance, but the real-time

requirements of cross-border forensics make it

difficult to guarantee procedural compliance, which

leads to the fuzziness of digital residences. At the

same time, there are also difficulties in the review of

the legitimacy of evidence. For example, when the

U.S. police use AI to generate a crime report, if the

data comes from a third country server, the defendant

often questions the validity of evidence by violating

the data sovereignty law, causing a crisis of mutual

trust in international justice.

2.2.2 Algorithm Discrimination Risk

Taking the controversy of predictive policing in the

United States as an example, the algorithm of

historical arrest data training (such as predpol) marks

the black community as a high-risk area, resulting in

excessive deployment of police force and screening

discrimination. Research conducted by the

Massachusetts Institute of Technology and the

National Institute of Standards and Technology

(NIST) shows that many face recognition

technologies in the United States have poor

recognition accuracy for people of color. At the same

time, many black people have encountered

algorithmic discrimination.

First, the application of AI in criminal

investigation will lead to the failure of transparency

mechanism. In some cases, the AI supplier will refuse

to disclose the algorithm logic on the grounds of trade

secret, which makes the defendant unable to

effectively cross-examine; The EU AI Act requires

algorithm interpretability, but technology companies

use zero knowledge proof and other tools to avoid it,

which essentially forms pseudo transparency.

In addition, there is a vacuum in the attribution of

responsibility. When AI decisions lead to false arrest

(such as the case of Robert Williams, a black man in

Detroit), the police will blame the algorithm defect,

while the developer invokes the user agreement

exemption clause, which forms a double

responsibility escape.

Regional governance attempts, such as the new

artificial intelligence act of the European Union,

which prohibits high-risk algorithms, but the

legislation of various states in the United States is

uneven (for example, California prohibits police face

recognition, while Texas encourages the use), leading

to the problem of compliance puzzle for multinational

enterprises.

At the same time, developing countries are still

dependent on technology. Due to backward

technology and insufficient innovation, some

countries' data services are difficult to have a foothold

in the market. In the long run, developed countries

almost take over the data services of some countries,

and the local bias of their algorithms is questioned,

which has the risk of penetrating into the data of other

countries, thus triggering the controversy of digital

colonization.

3 RULE DILEMMA OF CROSS

BORDER DATA GOVERNANCE

3.1 Defects in the Current Rule System

3.1.1 Lag of WTO Rules

The cross-border data governance rules under the

WTO framework still adhere to the traditional

consensus decision-making mechanism, which

cannot adapt to the rapid development of the digital

economy. This mechanism is inefficient in updating

rules and cannot meet the real-time needs of cross-

border data flow.

In addition, the existing WTO rules (such as the

general agreement on trade in services) do not

explicitly cover the specific standards of data cross-

border flow, and the dispute settlement body (DSB)

lacks the legal basis for handling digital trade

disputes. For example, the conflict between data

privacy protection and trade liberalization between

the United States and the European Union is difficult

to be resolved through WTO judicial channels due to

the lack of uniform substantive rules. The WTO

dispute settlement mechanism focuses more on

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

450

traditional trade disputes in goods and lacks

explanatory power on new issues such as data

sovereignty and algorithm transparency.

3.1.2 Extraterritorial Expansion of Domestic

Laws

The United States has implemented extraterritorial

jurisdiction through the cloud act and the chip and

Science Act, requiring enterprises to provide overseas

storage data, and even including Chinese technology

enterprises in the entity list to restrict technology

exports. Such unilateral measures lead to direct

conflicts with other countries' data sovereignty. For

example, after the European Court overturned the

safe harbor agreement, the EU built data flow barriers

through the general data protection regulations

(GDPR), forming a rule hedge.

In addition, countries have expanded data cross-

border regulation on the grounds of national security,

such as India's ban on Chinese applications and the

United States' exclusion of Huawei's equipment with

the clean network plan. Such measures are often

questioned as digital protectionism. For example, the

adequacy determination mechanism of the EU GDPR

has been criticized as a disguised data flow barrier,

leading to a surge in compliance costs for enterprises.

The extraterritorial expansion of domestic laws

will also trigger transnational legal confrontation.

Typical cases include the United States' request to

Canada to arrest Meng Wanzhou, a Huawei

executive, and so on. Such conflicts are difficult to

resolve due to the lack of international coordination

mechanisms. For example, the multiple criteria for

determining the connection points such as the place

of data storage and the nationality of the processing

subject exacerbate the jurisdictional disputes.

3.2 Core Contradiction Analysis

3.2.1 Contradiction Between Efficiency and

Safety

There are contradictions between the free flow of data

and localized storage. Cross-border data flow is the

core driving force for the development of digital

economy, but its natural boundlessness has

fundamental conflicts with national security and

personal privacy protection. For example, the

European Union has established strict rules for cross-

border transmission through the general data

protection regulations (GDPR), requiring data

receiving countries to pass adequacy identification to

prove that their protection level is equivalent to that

of the European Union (Voss, 2019). Although this

mechanism strengthens the protection of privacy, it

leads to high compliance costs for multinational

enterprises (such as the establishment of local data

centers), which hinders the efficiency of data flow.

The United States adopted the cloud act to implement

the data controller principle, allowing the government

to access enterprise data across borders to improve

law enforcement efficiency, but was criticized by the

European Union as sacrificing the sovereignty of

other countries with efficiency.

In addition, there is a symbiotic relationship

between technological empowerment and security

vulnerabilities. Although new technologies such as

blockchain and privacy computing can enhance the

credibility of cross-border data flow (such as "zero

knowledge proof" to make data available and

invisible), they may also be used to avoid sovereign

regulations. For example, the anonymity of the dark

net and cryptocurrency provides a channel for

transnational crime, forcing countries to make a

difficult trade-off between improving data tracking

ability and protecting citizens' privacy. At the same

time, the global layout of cloud computing services

weakens the relevance between the physical storage

place of data and jurisdiction, and the traditional

territorial principle faces the risk of failure.

3.2.2 Conflict Between Technology and

Sovereignty

At present, the core contradiction of cross-border data

governance has evolved from a simple legal conflict

to the competition between technical standards and

rule systems. The United States has passed the chip

and science act to restrict technology exports to

China. At the same time, the United States has

established a data flow alliance with its allies to turn

technological advantages into the right to speak on

rules; The construction of EU compliance barriers

requires that the privacy protection mechanism be

embedded in the AI system design stage, forcing non-

EU enterprises to restructure their technical

architecture to meet European standards; China's

promotion of localized data governance templates

through the belt and road initiative has been

questioned as digital rule output (Ozalp et al.,2022；

Luo& Van Assche,2023).

In addition, developed countries rely on their

technological advantages to form a data gravity

effect, leading to the passivity of developing

countries. At the same time, the output of algorithm

bias is also a serious problem, which is easy to cause

the controversy of digital colonization.

The Application of Artiﬁcial Intelligence in the Process of Intelligent Criminal Justice: From the Perspective of Cross-Border Data

Governance

451

4 IMPROVEMENT PATH OF

CROSS BORDER DATA

GOVERNANCE

4.1 Rule Innovation

4.1.1 Establish Investigation Data

Classification System

In cross-border data governance, it is necessary to

implement classified and hierarchical management

according to data sensitivity. For example, DNA,

Biometrics and other core data related to personal

privacy or national security should be strictly

prohibited from cross-border flow, while low-risk

data such as IP addresses and public transaction

records can be conditionally shared through the

negotiation mechanism. China's data security law has

put forward a hierarchical framework of core data -

important data - General data, which can be further

refined to specific scenarios, such as distinguishing

terrorism related and classified data from ordinary

case clue data in investigation data and clarifying

cross-border transmission rules at different levels.

4.1.2 Developing International Standards

for AI Investigation

For the application of AI in criminal investigation, it

is necessary to promote the unification of

international technical standards. For example, the

EU AI Act requires that high-risk algorithms need to

be embedded with interpretability and privacy

protection mechanisms, while China can work with

BRICs countries to develop AI Investigation

Technical specifications that take into account

efficiency and security, covering data desensitization,

algorithm transparency, evidence chain traceability

and other dimensions. The principle of human rights

centrism proposed by UNESCO can also provide an

ethical framework for global AI investigation

standards.

4.2 Mechanism Construction

4.2.1 Establishment of International Ai

Investigation Compliance Committee

Establish a multilateral institution composed of

sovereign states and technical legal scholars to review

the compliance of the AI investigation system. For

example, the risk of racial discrimination in

predictive policing algorithms and the legitimacy of

cross-border data call procedures are dynamically

evaluated, and innovative technology applications are

piloted through the regulatory sandbox mechanism.

4.2.2 Pilot Cross-Border Counter-Terrorism

Intelligence Exchange

The Sino Russian "border defense cooperation-2024"

joint anti-terrorism exercise has practiced the cross-

border intelligence sharing mechanism and realized

real-time data collaboration through technical means

such as air reconnaissance and water interception. In

the future, such models can be promoted, and

Regional Anti-Terrorism data exchange platforms

can be established under the frameworks of ASEAN

and the Shanghai Cooperation Organization to clarify

the scope of shared data (such as encrypted

communication metadata), authority (such as judicial

authorization) and dispute resolution rules.

4.3 China's Coping Strategies

4.3.1 Actively Participate in the

Formulation of International Rules

China needs to promote the concept of safe and

orderly flow of data into international agreements

such as CPTPP and DEPA on the basis of the global

cross border data flow Cooperation Initiative. For

example, the "negative list+security assessment

system" was piloted by RCEP to allow the free flow

of data that does not involve national security. At the

same time, the provisions on personal information

protection were improved against the EU GDPR to

enhance the compatibility of rules.

4.3.2 Improving Domestic Support Systems

China has refined the enforcement rules of the data

security law and the personal information protection

law for domestic demand, such as establishing a

"dynamic database for data exit security assessment”

and relying on blockchain technology to achieve the

full life cycle of cross-border data storage. At the

same time, we should cultivate professional cross-

border data service institutions, provide one-stop

support for enterprises such as compliance consulting

and risk assessment, and reduce the cost of going to

sea compliance. At the technical level, we will

increase investment in research and development of

technologies such as privacy computing and

homomorphic encryption and build an independent

and controllable infrastructure for cross-border data

flow.

ICPLSS 2025 - International Conference on Politics, Law, and Social Science

452

5 CONCLUSION

AI investigation has become an inevitable choice to

improve the efficiency of criminal justice through

crime prediction, evidence correlation, risk early

warning and other technical means. However, the

cross-border data flows it relies on (such as

biometrics and communication records)

fundamentally conflict with the existing international

rules: the localization requirements of the EU GDPR,

the extraterritorial jurisdiction of the US cloud act and

the data sovereignty demands of developing countries

form a structural contradiction. Data shows that 80%

of the world's data is stored on servers in the United

States and Europe, and developing countries are in a

passive position in the data value chain, highlighting

the urgency of rule reconstruction.

In the future, rules need to be innovated, such as

establishing the classification system of investigation

data (such as the prohibition of cross-border DNA

data and conditional sharing of IP addresses) and the

international standard of AI investigation, promoting

the mutual recognition of algorithm transparency

(such as the interpretability requirements of the EU

AI act) and data desensitization technology, and

reducing technical barriers.

Carry out relevant mechanism construction, such

as the establishment of the International Ai

investigation compliance committee, relying on the

UN framework to review the risk of algorithm

discrimination, and pilot the regulatory sandbox to

verify the technical compliance.

REFERENCES

Barbir, M. & Stankovic, B. 2024. Collecting and using

electronic evidence in criminal proceedings. In

Collection Papers from Conference Organized on

Occasion Day Faculty of Law 326.

Blount, K. 2024. Using artificial intelligence to prevent

crime: Implications for due process and criminal

justice. AI & Society 39(1): 359-368.

Cohen, J.N. 2024. Adapting to AI: How will generative AI

affect work? How should we respond? SocArXiv.

Dai, Y. 2023. Artificial intelligence regulation from the

perspective of international trade law: From the

perspective of WTO rules. Journal of Shanghai

University of Finance and Economics 25(2): 122-

136+152.

Li, H. 2020. The reform of criminal investigation work

mode in the era of artificial intelligence. China Criminal

Police (3): 9-12.

Luo, Y. & Van Assche, A. 2023. The rise of techno-

geopolitical uncertainty: Implications of the United

States CHIPS and Science Act. Journal of International

Business Studies.

Ozalp, H. et al. 2022. “Digital colonization” of highly

regulated industries: An analysis of big tech platforms’

entry into health care and education. California

Management Review 64(4): 78-107.

Re, R.M. & Solow-Niederman, A. 2019. Developing

artificially intelligent justice. Stanford Technology Law

Review 22: 242.

Voss, W.G. 2019. Cross-border data flows, the GDPR, and

data governance. Washington International Law

Journal 29: 485.

Zhang, Z., Zhang, J. & Tan, T. 2021. Current situation

analysis and countermeasures of artificial intelligence

ethics. Journal of the Chinese Academy of Sciences

36(11): 1270-1277.

The Application of Artiﬁcial Intelligence in the Process of Intelligent Criminal Justice: From the Perspective of Cross-Border Data

Governance

453