Introduction and Brief Analysis

A movie named “The Artifice Girl” portrayed A law enforcement agency developing an AI-based personification of a 12-year-old girl who appears to be exactly like a real person. Believing her to be an actual girl, perpetrators of child sexual exploitation were caught attempting to seek sexual favours. The movie showed how AI aided law enforcement, but the reality is that the emergence of Artificial Intelligence has posed numerous challenges in multiple directions. This example illustrates both the promise and the complexity of using AI in sensitive areas like law enforcement, where technological innovation must be carefully balanced with ethical and legal considerations.

Detection and Protection tools are constantly competing with technologies that generate content, automate grooming and challenge legal boundaries. Such technological advancements have provided enough ground for the proliferation of Child Sexual Exploitation and Abuse Material (CSEAM). Also known as child pornography under Section 2 (da) of Protection of Children from Sexual Offences Act, 2012, it defined it as - “means any visual depiction of sexually explicit conduct involving a child which includes a photograph, video, digital or computer-generated image indistinguishable from an actual child and image created, adapted, or modified, but appears to depict a child.”

Artificial Intelligence is a category of technologies that attempt to shape human thoughts and behaviours using input algorithms and datasets. Two Primary applications can be considered in the context of CSEAM: classifiers and content generators. Classifiers are programs that learn from large data sets, which may be labelled or unlabelled and further classify what is restricted or illegal. Whereas generative AI is also trained on large datasets, it uses that knowledge to create new things. Majority of current AI research related to AI for CSEAM is done by the use of Artificial neural networks (ANNs), a type of AI that can be trained to identify unusual connections between items (classification) and to generate unique combinations of items (e.g., elements of a picture) based on the training data used.

Current Legal Landscape

‍

The legal Landscape in terms of AI is yet unclear and evolving, with different nations trying to track the evolution of AI and develop laws. However, some laws directly address CSEAM. The International Centre for Missing and Exploited Children (ICMEC) combats Illegal sexual content involving children. They have a “Model Legislation” for setting recommended sanctions/sentencing. According to research performed in 2018, Illegal sexual content involving children is illegal in 118 of the 196 Interpol member states. This figure represents countries that have sufficient legislation in place to meet 4 or 5 of the 5 criteria defined by the ICMEC.

CSEAM in India can be reported on various portals like the ‘National Cyber Crime Reporting Portal’. Online crimes related to children, including CSEAM, can be reported to this portal by visiting cybercrime.gov.in. This portal allows anonymous reporting, automatic FIR registration and tracking of your complaint. ‘I4C Sahyog Portal’ is another platform managed by the Indian Cyber Crime Coordination Centre (I4C). This portal integrates with social media platforms.

The Indian legal front for AI is evolving and CSEAM is well addressed in Indian laws and through judicial pronouncements. The Supreme Court judgement on Alliance and Anr v S Harish and ors is a landmark in this regard. The following principles were highlighted in this judgment.

‍

• The term “child pornography” should be substituted by “Child Sexual Exploitation and Abuse Material” (CSEAM) and shall not be used for any further judicial proceeding, order, or judgment. Also, parliament should amend the same in POCSO and instead, the term CSEAM should be endorsed.
• Parliament to consider amending Section 15 (1) of POCSO to make it more convenient for the general public to report by way of an online portal.
• Implementing sex education programs to give young people a clear understanding of consent and the consequences of exploitation. To help prevent Problematic sexual behaviour (PSB), schools should teach students about consent, healthy relationships and appropriate behaviour.
• Support services to the victims and rehabilitation programs for the offenders are essential.
• Early identification of at-risk individuals and implementation of intervention strategies for youth.

‍

Distinctive Challenges

According to a report by the National Centre for Missing and Exploited Children (NCMEC), a significant number of reports about child sexual exploitation and abuse material (CSEAM) are linked to perpetrators based outside the country. This highlights major challenges related to jurisdiction and anonymity in addressing such crimes. Since the issue concerns children and considering the cross-border nature of the internet and the emergence of AI, Nations across the globe need to come together to solve this matter. Delays in the extradition procedure and irregular legal processes across the jurisdictions hinder the apprehension of offenders and the delivery of justice to victims.

CyberPeace Recommendations

For effective regulation of AI-generated CSEAM, laws are required to be strengthened for AI developers and trainers to prevent misuse of their tools. AI should be designed with its ethical considerations, ensuring respect for privacy, consent and child rights. There can be a self-regulation mechanism for AI models to recognise and restrict red flags related to CSEAM and indicate grooming or potential abuse.

A distinct Indian CSEAM reporting portal is urgently needed, as cybercrimes are increasing throughout the nation. Depending on the integrated portal may lead to ignorance of AI-based CSEAM cases. This would result in faster response and focused tracking. Since AI-generated content is detectable. The portal should also include an automated AI-content detection system linked directly to law enforcement for swift action.

Furthermore, International cooperation is of utmost importance to win the battle of AI-enabled challenges and to fill the jurisdictional gaps. A united global effort is required. Using a common technology and unified international laws is essential to tackle AI-driven child sexual exploitation across borders and protect children everywhere. CSEAM is an extremely serious issue. Children are among the most vulnerable to such harmful content. This threat must be addressed without delay, through stronger policies, dedicated reporting mechanisms and swift action to protect children from exploitation.

‍

References:

• https://www.sciencedirect.com/science/article/pii/S2950193824000433?ref=pdf_download&fr=RR-2&rr=94efffff09e95975
• https://aasc.assam.gov.in/sites/default/files/swf_utility_folder/departments/aasc_webcomindia_org_oi d_4/portlet/level_2/pocso_act.pdf
• https://www.manupatracademy.com/assets/pdf/legalpost/just-rights-for-children-alliance-and-anr-vs-s‍harish-and-ors.pdfhttps://www.icmec.org‍https://www.missingkids.org/theissues/generative-ai

‍

Executive Summary:

Traditional Business Email Compromise(BEC) attacks have become smarter, using advanced technologies to enhance their capability. Another such technology which is on the rise is WormGPT, which is a generative AI tool that is being leveraged by the cybercriminals for the purpose of BEC. This research aims at discussing WormGPT and its features as well as the risks associated with the application of the WormGPT in criminal activities. The purpose is to give a general overview of how WormGPT is involved in BEC attacks and give some advice on how to prevent it.

 Introduction

BEC(Business Email Compromise) in simple terms can be defined as a kind of cybercrime whereby the attackers target the business in an effort to defraud through the use of emails. Earlier on, BEC attacks were executed through simple email scams and phishing. However, in recent days due to the advancement of AI tools like WormGPT such malicious activities have become sophisticated and difficult to identify. This paper seeks to discuss WormGPT, a generative artificial intelligence, and how it is used in the BEC attacks to make the attacks more effective.

What is WormGPT?

Definition and Overview

WormGPT is a generative AI model designed to create human-like text. It is built on advanced machine learning algorithms, specifically leveraging large language models (LLMs). These models are trained on vast amounts of text data to generate coherent and contextually relevant content. WormGPT is notable for its ability to produce highly convincing and personalised email content, making it a potent tool in the hands of cybercriminals.

How WormGPT Works

1. Training Data: Here the WormGPT is trained with the arrays of data sets, like emails, articles, and other writing material. This extensive training enables it to understand and to mimic different writing styles and recognizable textual content.

 2. Generative Capabilities: Upon training, WormGPT can then generate text based on specific prompts, as in the following examples in response to prompts. For example, if a cybercriminal comes up with a prompt concerning the company’s financial information, WormGPT is capable of releasing an appearance of a genuine email asking for more details.

 3. Customization: WormGPT can be retrained any time with an industry or an organisation of interest in mind. This customization enables the attackers to make their emails resemble the business activities of the target thus enhancing the chances for an attack to succeed.

Enhanced Phishing Techniques

Traditional phishing emails are often identifiable by their generic and unconvincing content. WormGPT improves upon this by generating highly personalised and contextually accurate emails. This personalization makes it harder for recipients to identify malicious intent.

Automation of Email Crafting

Previously, creating convincing phishing emails required significant manual effort. WormGPT automates this process, allowing attackers to generate large volumes of realistic emails quickly. This automation increases the scale and frequency of BEC attacks.

Exploitation of Contextual Information

WormGPT can be fed with contextual information about the target, such as recent company news or employee details. This capability enables the generation of emails that appear highly relevant and urgent, further deceiving recipients into taking harmful actions.

Implications for Cybersecurity

Challenges in Detection

The use of WormGPT complicates the detection of BEC attacks. Traditional email security solutions may struggle to identify malicious emails generated by advanced AI, as they can closely mimic legitimate correspondence. This necessitates the development of more sophisticated detection mechanisms.

Need for Enhanced Training

Organisations must invest in training their employees to recognize signs of BEC attacks. Awareness programs should emphasise the importance of verifying email requests for sensitive information, especially when such requests come from unfamiliar or unexpected sources.

Implementation of Robust Security Measures

1. Multi-Factor Authentication (MFA): MFA can add an additional layer of security, making it harder for attackers to gain unauthorised access even if they successfully deceive an employee.
2. Email Filtering Solutions: Advanced email filtering solutions that use AI and machine learning to detect anomalies and suspicious patterns can help identify and block malicious emails.
3. Regular Security Audits: Conducting regular security audits can help identify vulnerabilities and ensure that security measures are up to date.

Case Studies

Case Study 1: Financial Institution

A financial institution fell victim to a BEC attack orchestrated using WormGPT. The attacker used the tool to craft a convincing email that appeared to come from the institution’s CEO, requesting a large wire transfer. The email’s convincing nature led to the transfer of funds before the scam was discovered.

Case Study 2: Manufacturing Company

In another instance, a manufacturing company was targeted by a BEC attack using WormGPT. The attacker generated emails that appeared to come from a key supplier, requesting sensitive business information. The attack exploited the company’s lack of awareness about BEC threats, resulting in a significant data breach.

Recommendations for Mitigation

1. Strengthen Email Security Protocols: Implement advanced email security solutions that incorporate AI-driven threat detection.
2. Promote Cyber Hygiene: Educate employees on recognizing phishing attempts and practising safe email habits.
3. Invest in AI for Defense: Explore the use of AI and machine learning in developing defences against generative AI-driven attacks.
4. Implement Verification Procedures: Establish procedures for verifying the authenticity of sensitive requests, especially those received via email.

Conclusion

WormGPT is a new tool in the arsenal of cybercriminals which improved their options to perform Business Email Compromise attacks more effectively and effectively. Therefore, it is critical to provide the defence community with information regarding the potential of WormGPT and its implications for enhancing the threat landscape and strengthening the protection systems against advanced and constantly evolving threats. 

This means the development of rigorous security protocols, general awareness of security solutions, and incorporating technologies such as artificial intelligence to mitigate the risk factors that arise from generative AI tools to the best extent possible.

Introduction

In today’s digital world, where everything is related to data, the more data you own, the more control and compliance you have over the market, which is why companies are looking for ways to use data to improve their business. But at the same time, they have to make sure they are protecting people’s privacy. It is very tricky to strike a balance between both of them. Imagine you are trying to bake a cake where you need to use all the ingredients to make it taste great, but you also have to make sure no one can tell what’s in it. That’s kind of what companies are dealing with when it comes to data. Here, ‘Pseudonymisation’ emerges as a critical technical and legal mechanism that offers a middle ground between data anonymisation and unrestricted data processing. 

Legal Framework and Regulatory Landscape 

Pseudonymisation, as defined by the General Data Protection Regulation (GDPR) in Article 4(5), refers to “the processing of personal data in such a manner that the personal data can no longer be attributed to a specific data subject without the use of additional information, provided that such additional information is kept separately and is subject to technical and organisational measures to ensure that the personal data are not attributed to an identified or identifiable natural person”. This technique represents a paradigm shift in data protection strategy, enabling organisations to preserve data utility while significantly reducing privacy risks. The growing importance of this balance is evident in the proliferation of data protection laws worldwide, from GDPR in Europe to India’s Digital Personal Data Protection Act (DPDP) of 2023. 

Its legal treatment varies across jurisdictions, but a convergent approach is emerging that recognises its value as a data protection safeguard while maintaining that the pseudonymised data remains personal data. Article 25(1) of GDPR recognises it as “an appropriate technical and organisational measure” and emphasises its role in reducing risks to data subjects. It protects personal data by reducing the risk of identifying individuals during data processing. The European Data Protection Board’s (EDPB) 2025 Guidelines on Pseudonymisation provide detailed guidance emphasising the importance of defining the “pseudonymisation domain”. It defines who is prevented from attributing data to specific individuals and ensures that the technical and organised measures are in place to block unauthorised linkage of pseudonymised data to the original data subjects. In India, while the DPDP Act does not explicitly define pseudonymisation, legal scholars argue that such data would still fall under the definition of personal data, as it remains potentially identifiable. The Act defines personal data defined in section 2(t) broadly as “any data about an individual who is identifiable by or in relation to such data,” suggesting that the pseudonymised information, being reversible, would continue to require compliance with data protection obligations. 

Further, the DPDP  Act, 2023 also includes principles of data minimisation and purpose limitation. Section 8(4) says that a “Data Fiduciary shall implement appropriate technical and organisational measures to ensure effective observance of the provisions of this Act and the Rules made under it.” The concept of Pseudonymization fits here because it is a recognised technical safeguard, which means companies can use pseudonymization as one of the methods or part of their compliance toolkit under Section 8(4) of the DPDP Act. However, its use should be assessed on a case to case basis, since ‘encryption’ is also considered one of the strongest methods for protecting personal data. The suitability of pseudonymization depends on the nature of the processing activity, the type of data involved, and the level of risk that needs to be mitigated. In practice, organisations may use pseudonymization in combination with other safeguards to strengthen overall compliance and security.

The European Court of Justice’s recent jurisprudence has introduced nuanced considerations about when pseudonymised data might not constitute personal data for certain entities. In cases where only the original controller possesses the means to re-identify individuals, third parties processing such data may not be subject to the full scope of data protection obligations, provided they cannot reasonably identify the data subjects. The “means reasonably likely” assessment represents a significant development in understanding the boundaries of data protection law. 

Corporate Implementation Strategies 

Companies find that pseudonymisation is not just about following rules, but it also brings real benefits. By using this technique, businesses can keep their data more secure and reduce the damage in the event of a breach. Customers feel more confident knowing that their information is protected, which builds trust. Additionally, companies can utilise this data for their research or other important purposes without compromising user privacy. 

Key Benefits of Pseudonymisation:

1. Enhanced Privacy Protection: It hides personal details like names or IDs with fake ones (with artificial values or codes), making it harder for accidental privacy breaches.
2. Preserved Data Utility: Unlike completely anonymous data, pseudonymised data keeps its usefulness by maintaining important patterns and relationships within datasets.
3. Facilitate Data Sharing: It’s easier to share pseudonymised data with partners or researchers because it protects privacy while still being useful.

However, using pseudonymisation is not as easy as companies have to deal with tricky technical issues like choosing the right methods, such as encryption or tokenisation and managing security keys safely. They have to implement strong policies to stop anyone from figuring out who the data belongs to. This can get expensive and complicated, especially when dealing with a large amount of data, and it often requires expert help and regular upkeep. 

Balancing Privacy Rights and Data Utility

The primary challenge in pseudonymisation is striking the right balance between protecting individuals' privacy and maintaining the utility of the data. To get this right, companies need to consider several factors, such as why they are using the data, the potential hacker's level of skill, and the type of data being used.

Conclusion 

Pseudonymisation offers a practical middle ground between full anonymisation and restricted data use, enabling organisations to harness the value of data while protecting individual privacy. Legally, it is recognised as a safeguard but still treated as personal data, requiring compliance under frameworks like GDPR and India’s DPDP Act. For companies, it is not only regulatory adherence but also ensuring that it builds trust and enhances data security. However, its effectiveness depends on robust technical methods, governance, and vigilance. Striking the right balance between privacy and data utility is crucial for sustainable, ethical, and innovation-driven data practices.

References:

1. https://gdpr-info.eu/art-4-gdpr/
2. https://www.meity.gov.in/static/uploads/2024/06/2bf1f0e9f04e6fb4f8fef35e82c42aa5.pdf
3. https://gdpr-info.eu/art-25-gdpr/
4. https://www.edpb.europa.eu/system/files/2025-01/edpb_guidelines_202501_pseudonymisation_en.pdf
5. https://curia.europa.eu/juris/document/document.jsf?text=&docid=303863&pageIndex=0&doclang=EN&mode=req&dir=&occ=first&part=1&cid=16466915
6. https://curia.europa.eu/juris/document/document.jsf?text=&docid=303863&pageIndex=0&doclang=EN&mode=req&dir=&occ=first&part=1&cid=16466915

‍