AI Image Categorization and Text Analytics Detect Narcotics Market on Social Media
When startups such as Facebook, Twitter and LinkedIn began their quest to dominate the social media landscape they did not consider the need to “police” their digital arena for criminal activity. Now that they account for almost 3.5 billion users, the immense responsibility of protecting their users by vetting criminal activity has become critical to the safety of their communities.
As public concern escalates over the unintentional hosting of explicit drug trafficking, social media networks are being called out to ‘up their game’ and ensure a safer digital environment.
While their initial emphasis was getting as many users as possible and monetizing them through advertising, it has become clear that these same communication channels have been exploited by drug dealers.
Now that the battle for market share has stabilized, the time has come for social networks to apply their cutting-edge technology in ways that deliver more than just “neighborhood watch” policies that encourage users to flag offensive content or alert support. But the sheer speed and volume of interactions on their platforms has ruled out human oversight as the agile drug market constantly evolves tactics to evade detection.
In order to contend with the current velocity and volume of digital interactions, Artificial Intelligence and Machine Learning has risen to the occasion, promising to scale real time monitoring of threats, while anticipating dangers that criminal elements and bad actors pose.
Facebook and Instagram fight back
Facebook has begun to go beyond relying on users to report drug-related images and content. The world’s biggest social media platform has been deploying Artificial Intelligence to spot both drug content and calculate who the drug dealers are on the network.
As the US opioid crisis has demonstrated, drug dealers and buyers welcome social media as a convenient arena for the narcotics market. However, with recent drug-related image categorization capabilities, Facebook’s AI is able to ‘detect and reject’ content that explicitly or implicitly promotes the selling or buying of illicit drugs.
“Our technology is able to detect content that includes images of drugs and depicts the intent to sell with information such as price, phone numbers or usernames for other social media accounts,” said Kevin Martin, head of US public policy at Facebook. – source
Facebook is not the only social network trying to curb illicit drug and prescription trafficking on its platform.
Twitter becomes proactive
Twitter has launched initiatives to ‘scan and ban’ drug tweeting content while monitoring hashtags. Hashtags were created to assign a theme to a particular tweet, allowing thematic posts to be easily found throughout the platform. Hashtags are comprised of letters, digits, and underscores preceded by the hash symbol, #.
“We have an iterative, proactive process in place to ensure that we prevent opportunities for – and respond quickly to – illicit drug sales on our platforms. We are also working closely with experts to take all possible actions and have explicit policies in place that help make our platform safe for our community.” Twitter spokesperson – source
Some of the hashtags indicating narcotics availability and prices are explicit (#BuyDrugsOnline), some are found under street-names (“Captain Cody” = Codeine) while others are indicated by cryptic hashtags (#white = fentanyl). With half a billion messages posted per day, staying on top of drug content censorship is a massive task.
LinkedIn must start to monitor company profiles
Even LinkedIn has shown signs of becoming a haven for illegal online pharmacies that sell Adderall, Xanax, Vicodin and more. With certain “companies” going by dubious names such as “Adderall For Sale“, a wider threat to social media users is going undetected. When 76% of US citizens* are likely to buy medicine online, the proliferation of illegally operated pharmacies can pose a national threat to the health of Americans and the wider global community.
(*according to a recent survey from the nonprofit Alliance for Safe Online Pharmacies)
AI-powered detection solutions
The list of social networks where illicit and controlled substances are being sold continues to grow and now also includes Tumblr, Pinterest, Google+, WhatsApp, Reddit, and more. Because of this expansion of drug trafficking activity, it has become increasingly clear that these other social media hubs will also need to implement an AI solution that will both alert authorities to the illegal sale of controlled substances online and remove offending accounts.
“What we need to do is build more AI tools that can proactively find that (drug-related) content.” – Facebook CEO, Mark Zuckerberg
With the recent introduction of “proactive detection” powered by AI, Facebook and Instagram are demonstrating how to more effectively monitor drug trafficking in real time while freeing up human moderators to focus on pages, groups and hashtags.
Social media giants working together
Facebook, Twitter and Google recently formed Tech Together to Fight the Opioid Crisis in an effort to both combat the US opioid crisis and understand how to deal with ongoing illicit drug trafficking on their platforms.
With the assistance of the University of Alabama’s Computer Forensic Research Lab, bad actors and their content on social networks are now being flagged and dealt with. Moreover, when users are identified as struggling with addiction, algorithms serve them content that directs them to confidential treatment and education.
Cellebrite AI-powered Analytics identifies opioid dealer
Cellebrite has also recognized the need for AI image categorization as drug-related content continues to surface in digital investigations. With Cellebrite Pathfinder, predefined categories as well as customized categories have been introduced to more effectively identify drugs, weapons and money as well as many more objects that could indicate crime-related activity.
In a recent case, Cellebrite Pathfinder successfully analyzed digital evidence extracted from 41 phones to find a common opioid dealer now serving his sentence for murder. To understand more examples of Analytics in action, read the full case study here, or view the webinar, “Opioid Crisis in America: From Digital Clues to a Murder Conviction.”