Software solutions to identify users behind Telegram, WhatsApp and Instagram based drug trafficking
Abstract : Background: Use of encrypted messaging/social media apps like Telegram, WhatsApp and Instagram for drug trafficking are on the rise. Channels operating on Telegram and WhatsApp and Instagram handles are blatantly being misused by drug traffickers for offering various narcotic drugs and Psychotropic substances for sale. Description: WhatsApp and Telegram channels and Instagram handles are created by drug traffickers to offer various drugs for sale to their subscribers. Customized Telegram bots are also created by some of the drug traffickers to sell drugs.
It is most worrisome that majority of the drugs which are being offered on sale through Telegram, WhatsApp and Instagram are dangerous synthetic drugs like MDMA, LSD, Mephedrone etc. The above three apps are also used by drug traffickers for drug communication. Expected Solution: Development of a software solution to identify live Telegram and WhatsApp channels/bots and Instagram handles that are offering drugs for sale in India. Solution also should focus on triangulating identifiable parameters like IP address, mobile number, email id etc of the users behind the channel/bot/handle.
Since the first documentation of app- and social-media-based drug dealing in 2018, there has been a growing debate on these activities in the media and in prevention and law enforcement circles. This is likely to have led to stricter platform moderation and policing initiatives.
Hence, increased moderation might have reversed the move to open social media, making drug markets more reliant on one-to-one contact and likely to be limited to the buyer’s network.
Based on these two interlinked hypotheses, indeed, we expect that social media markets for illicit drugs have moved away from the more open style of dealing back towards a hybrid open and locally embedded closed style. The focus of this report is to understand the current state of social media drug markets and how the drugs market has changed over the last five years.
We understand hybrid markets as drug purchases that are partially made online, but which, importantly, include some physical locality when meeting to exchange the drugs and money. Therefore, this report will not engage with purely online national or international markets where payment and drug delivery are achieved without meeting. An example could be Google groups, where internationally shipped drugs are advertised or dark-web-based markets (darknet markets)
End-to-End Encryption
Limited Access: Many messaging platforms, including WhatsApp and Telegram (in secret chat mode), use end-to-end encryption to secure messages. This means that even if data is intercepted, it is not readable without the encryption keys. Law enforcement agencies and other entities might struggle to access or decipher this data.
Anonymity and Privacy
Challenges in Tracing: Users often use pseudonyms or anonymous accounts, which makes it difficult to trace their real identities. Privacy features and anonymizing tools can further complicate tracking efforts.
Volume of Data
Overwhelming Amount of Information: Platforms generate vast amounts of data daily. Sifting through this information to find relevant clues related to drug trafficking can be overwhelming and time-consuming.
In this work, we propose to tackle the problem of drug dealer identification by constructing a large-scale dataset called Identifying Drug Dealers on Instagram (IDDIG), including over 2,000 posts, nearly 4,000 user homepages as well as multimodal data sources (text and images, posts and homepage).
In particular, the importance of biography information at the homepage on social media data mining has remained an underexplored topic [30, 35] and not been studied in previous works on drug-dealer detection [44, 76].
To construct such a large-scale dataset, we have designed an automatic data crawling system for Instagram that jointly uses hashtag and image information to guide the data collection. A user-friendly data annotation platform is also developed to support manual labeling of multimodal data, which is inevitably a tedious process.
With labeled data (ground-truth for machine learning), we have built a brand new drug-dealer identification system by leveraging the latest advances in deep learning including Bidirectional Encoder Representations from Transformers (BERT) [10] based text classification, ResNet-based [28] image classification, as well as feature-level multimodal data fusion [16]
Enhanced Detection Capabilities
Analysis: Advanced software tools use machine learning and AI to scan large volumes of data for suspicious patterns, keywords, and behaviors. This automation allows for faster detection of potential drug trafficking activities compared to manual methods.
Integration and Cross-Referencing
Data Correlation: Software solutions can integrate data from multiple sources and platforms, enabling cross-referencing and identifying connections between different users and activities. This can help in uncovering larger trafficking networks that span across various communication channels.
Real-Time Monitoring
Immediate Alerts: Some tools provide real-time monitoring and alert systems for detecting and flagging potentially illegal content or behaviors. This immediate response capability helps in taking swift action against suspected traffickers.
Content Analysis
Multimedia Detection: Modern software can analyze not only text but also images, videos, and audio files for signs of illegal activities. For instance, it can identify drug-related imagery or detect drug-related discussions within multimedia content.
|