Kirsten Chapman, Garrett Smith, Kaitlyn Klabacka, Joseph Thomas Bills, Addisyn Bushman, Terisa Gabrielsen, Pamela J. Wisniewski, Xinru Page
7 min
Young autistic adults often turn to social media for vital social connections, support networks, and practical help—like finding transportation or jobs—that are harder to access offline. However, they face heightened privacy risks due to how they uniquely interpret social media interfaces. This paper introduces PRISM (Privacy Rules for Inclusive Social Media), a 14-week classroom program designed specifically for autistic young adults with level 2 support needs (requiring substantial assistance). It evaluates whether tailored education can shift their privacy behaviors from extreme 'all-or-nothing' strategies to nuanced, safer practices.
Autistic young adults perceive social media 'affordances'—UI cues like 'friends' lists or share buttons—differently from neurotypical users. For instance, they might see 'friends' as inherently trustworthy, leading to oversharing with bad actors, cyberbullying, or physical harms. Prior studies show they experience disproportionate privacy violations, yet lack self-regulation skills to mitigate risks without completely avoiding platforms. This stems from neurodevelopmental differences, not ignorance, highlighting a gap in CSCW/HCI research: no prior empirically tested privacy education targeted this group, unlike interventions for children or elders.
PRISM leverages autistic individuals' strengths in rule-following and literal thinking. Instead of abstract lectures, it uses scenario-driven, contextual rules—realistic vignettes on platforms like Facebook and Instagram (e.g., 'If a new friend requests personal details immediately, apply Rule X: Decline and report'). Delivered in a classroom with 29 participants, it covered 6 topics like trust signals and sharing decisions. This contrasts with general-population training, affirming neurodiversity by matching communication styles and avoiding overload.
Pre- and post-assessments per topic showed statistically significant gains in safer decision-making. Participants moved from binary approaches (e.g., 'Quit social media entirely') to granular rules (e.g., 'Share location only with verified locals'). Engagement mattered: higher verbal contributions and peer interactions correlated with better outcomes, suggesting classroom dynamics amplify learning.
PRISM proves neuro-affirming interventions work, offering a blueprint for privacy educators: prioritize rules over intuition, scenarios over theory. For tech designers, it urges clearer affordance cues (e.g., explicit risk warnings) and supporter mediation tools. This fills a critical gap, enabling autistic users to safely reap social media's benefits without harms, with calls for scalable versions and longitudinal studies.
Young autistic adults may garner benefits through social media but also disproportionately experience privacy harms. Prior research found that these harms often stem from perceiving the affordances of social media differently than the general population, leading to unintentional risky behaviors and interactions with others. While educational interventions have been shown to increase social media privacy literacy for the general population, research has yet to focus on effective educational interventions for autistic young adults. We address this gap by developing and deploying Privacy Rules for Inclusive Social Media (PRISM), a classroom-based educational intervention tailored to the unique risks and neurodevelopmental differences of this population. Twenty-nine autistic students with substantial (level 2) support needs participated in a 14-week social media privacy literacy class. During these classes, participants often communicated their existing rule-based "all or nothing" approaches to privacy management (such as completely disengaging from social media to avoid privacy issues). Our course focused on empowering them by providing more nuanced guidance on safe privacy practices through the use of scenario-based formats and contextual, rule-based scenarios. Using pre- and post-knowledge assessments for each of our 6 course topics, our intervention led to a statistically significant increase in their making safer social media privacy decisions. We conclude with recommendations for how privacy educators and technology designers can leverage neuro-affirming educational interventions to increase privacy literacy for autistic social media users.
Alex: That sounds practical—like a cheat sheet for tricky online spots. And it moves them from black-and-white choices to checking context?
Sam: Precisely. Classes follow the same routine every week: quick start quiz, story and rule practice with group talk, then end quiz on the same questions. This predictability cuts mental overload, using visuals and easy words at about fourth-grade level. Across six topics—from platform basics to safe chats—the paper notes clearer gains in privacy smarts, with more nuanced picks post-class. Engagement mattered too; those chatting more or joining discussions showed stronger shifts.
Alex: So the structure itself helps learning stick, especially linking talk to better outcomes.
Sam: Yes. It suggests neuro-affirming setups—not fighting natural thinking styles—offer a solid path for safer online steps in this group. The evidence points to real, measurable progress without forcing conformity.
Alex: Okay, so the routine and checklists build skills module by module. But how did they actually check if it worked?
Sam: They gave everyone the same short quiz right before and after each lesson—multiple choice about real-life scenarios, like whether to share an email with a peer. This setup lets you see if a person remembers and applies the rules better immediately after, without relying on long-term memory, which can be tougher for this group. They scored each quiz as a fraction correct, then compared every student's own before and after using stats tests suited to the data. They skipped a control group because withholding helpful lessons from people at higher risk felt unethical.
Alex: So it's each person against themselves. Walk me through one of the modules, like sorting people into roles—how does that fit the rule idea?
Sam: One module builds on grouping people offline, like close family or teachers, but adds tags for online life—things like "online friends" or "strangers," which you can mix on one profile. Students practice labeling Facebook accounts this way, then learn safe sharing rules per tag: general info like age range okay for peers, but never passwords or home details for anyone. These rules turn vague feelings into checklists, matching strengths in patterns over gut instincts.
Alex: So tagging and safe-sharing rules prevent mix-ups, like oversharing with a "stranger" tagged as a peer.
Sam: Exactly. The study links stronger quiz jumps to class talk—those sharing thoughts or debating scenarios improved more, suggesting group practice reinforces the rules. In discussions, students shared strong views on privacy: some saw dangers everywhere, like scammers using fake profiles, while others overshared without worry or trusted all followers.
Alex: That coding of what they said—did it link back to who improved most on those quizzes?
Sam: It did. Those who spoke up more showed bigger quiz gains. Four out of six modules showed statistically significant score jumps—meaning the improvements were unlikely just from luck. For example, in sorting people into online roles like peers or strangers, scores rose from about half right to over 60% correct.
Alex: So some topics hit a sweet spot for rule practice, and those extra questions and shares signaled deeper thinking, tying to better scores.
Sam: Exactly. It underscores neuro-affirming teaching—letting natural ways of joining in lead to solid privacy steps. The paper suggests embracing autistic communication styles ties directly to stronger learning.
Alex: Right, so the teaching matched their strengths, leading to real shifts. But to weigh this fairly, what limits the strength of these findings?
Sam: A key limit is the small group of 29 students, with some missing sessions, which cuts statistical power. Assessments were short-term, right before and after each class, so they show immediate pickup but not if skills stick weeks later or change real behaviors. No control group was used, as skipping helpful lessons felt wrong ethically; plus, it focused on Facebook and Instagram only.
Alex: So solid first evidence, but needs scaling to confirm lasting impact.
Sam: Exactly. Practically, this points to apps delivering these rules asynchronously—letting users practice scenarios at their pace. Social media could add clearer risk warnings and step-by-step privacy sliders, helping nuanced choices without extremes. The paper positions PRISM as a foundation for inclusive tools, balancing safety with social benefits for autistic users.
Alex: That feels like a grounded path forward—tailored education plus smarter designs to cut harms while keeping connections open. Thanks, Sam—that's a clear look at how thoughtful teaching can make online spaces safer. Thanks for joining us on ResearchPod.