skip to main content

Title: Getting Meta: A Multimodal Approach for Detecting Unsafe Conversations within Instagram Direct Messages of Youth
Instagram, one of the most popular social media platforms among youth, has recently come under scrutiny for potentially being harmful to the safety and well-being of our younger generations. Automated approaches for risk detection may be one way to help mitigate some of these risks if such algorithms are both accurate and contextual to the types of online harms youth face on social media platforms. However, the imminent switch by Instagram to end-to-end encryption for private conversations will limit the type of data that will be available to the platform to detect and mitigate such risks. In this paper, we investigate which indicators are most helpful in automatically detecting risk in Instagram private conversations, with an eye on high-level metadata, which will still be available in the scenario of end-to-end encryption. Toward this end, we collected Instagram data from 172 youth (ages 13-21) and asked them to identify private message conversations that made them feel uncomfortable or unsafe. Our participants risk-flagged 28,725 conversations that contained 4,181,970 direct messages, including textual posts and images. Based on this rich and multimodal dataset, we tested multiple feature sets (metadata, linguistic cues, and image features) and trained classifiers to detect risky conversations. Overall, we found that the metadata features (e.g., conversation length, a proxy for participant engagement) were the best predictors of risky conversations. However, for distinguishing between risk types, the different linguistic and media cues were the best predictors. Based on our findings, we provide design implications for AI risk detection systems in the presence of end-to-end encryption. More broadly, our work contributes to the literature on adolescent online safety by moving toward more robust solutions for risk detection that directly takes into account the lived risk experiences of youth.  more » « less
Award ID(s):
1827700 2333207 1942610
Author(s) / Creator(s):
; ; ; ; ; ; ;
Date Published:
Journal Name:
Proceedings of the ACM on Human-Computer Interaction
Page Range / eLocation ID:
1 to 30
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. We collected Instagram data from 150 adolescents (ages 13-21) that included 15,547 private message conversations of which 326 conversations were flagged as sexually risky by participants. Based on this data, we leveraged a human-centered machine learning approach to create sexual risk detection classifiers for youth social media conversations. Our Convolutional Neural Network (CNN) and Random Forest models outperformed in identifying sexual risks at the conversation-level (AUC=0.88), and CNN outperformed at the message-level (AUC=0.85). We also trained classifiers to detect the severity risk level (i.e., safe, low, medium-high) of a given message with CNN outperforming other models (AUC=0.88). A feature analysis yielded deeper insights into patterns found within sexually safe versus unsafe conversations. We found that contextual features (e.g., age, gender, and relationship type) and Linguistic Inquiry and Word Count (LIWC) contributed the most for accurately detecting sexual conversations that made youth feel uncomfortable or unsafe. Our analysis provides insights into the important factors and contextual features that enhance automated detection of sexual risks within youths' private conversations. As such, we make valuable contributions to the computational risk detection and adolescent online safety literature through our human-centered approach of collecting and ground truth coding private social media conversations of youth for the purpose of risk classification. 
    more » « less
  2. Current youth online safety and risk detection solutions are mostly geared toward parental control. As HCI researchers, we acknowledge the importance of leveraging a youth-centered approach when building Artificial Intelligence (AI) tools for adolescents online safety. Therefore, we built the MOSafely, Is that ‘Sus’ (youth slang for suspicious)? a web-based risk detection assessment dashboard for youth (ages 13-21) to assess the AI risks identified within their online interactions (Instagram and Twitter Private conversations). This demonstration will showcase our novel system that embedded risk detection algorithms for youth evaluations and adopted the human–in–the loop approach for using youth evaluations to enhance the quality of machine learning models. 
    more » « less
  3. With the growing ubiquity of the Internet and access to media-based social media platforms, the risks associated with media content sharing on social media and the need for safety measures against such risks have grown paramount. At the same time, risk is highly contextualized, especially when it comes to media content youth share privately on social media. In this work, we conducted qualitative content analyses on risky media content flagged by youth participants and research assistants of similar ages to explore contextual dimensions of youth online risks. The contextual risk dimensions were then used to inform semi- and self-supervised state-of-the-art vision transformers to automate the process of identifying risky images shared by youth. We found that vision transformers are capable of learning complex image features for use in automated risk detection and classification. The results of our study serve as a foundation for designing contextualized and youth-centered machine-learning methods for automated online risk detection. 
    more » « less
  4. We collected Instagram Direct Messages (DMs) from 100 adolescents and young adults (ages 13-21) who then flagged their own conversations as safe or unsafe. We performed a mixed-method analysis of the media files shared privately in these conversations to gain human-centered insights into the risky interactions experienced by youth. Unsafe conversations ranged from unwanted sexual solicitations to mental health related concerns, and images shared in unsafe conversations tended to be of people and convey negative emotions, while those shared in regular conversations more often conveyed positive emotions and contained objects. Further, unsafe conversations were significantly shorter, suggesting that youth disengaged when they felt unsafe. Our work uncovers salient characteristics of safe and unsafe media shared in private conversations and provides the foundation to develop automated systems for online risk detection and mitigation. 
    more » « less
  5. In this work, we present a case study on an Instagram Data Donation (IGDD) project, which is a user study and web-based platform for youth (ages 13-21) to donate and annotate their Instagram data with the goal of improving adolescent online safety. We employed human-centered design principles to create an ecologically valid dataset that will be utilized to provide insights from teens’ private social media interactions and train machine learning models to detect online risks. Our work provides practical insights and implications for Human-Computer Interaction (HCI) researchers that collect and study social media data to address sensitive problems relating to societal good. 
    more » « less