Content moderators—who remove online content that violates social media guidelines—are repeatedly exposed to distressing material, making them vulnerable to anxiety, depression, and PTSD symptoms (e.g., intrusions). However, little research exists on harm reduction strategies for moderators. One proposed intervention, imagining bulletproof glass between themselves and harmful content, lacks empirical support. In this study, 202 participants completed a content moderator simulation, with half instructed to imagine bulletproof glass between themselves and the content throughout. All participants recorded and rated problematic characteristics of their image-related intrusions and completed pre- and post-task measures of state anxiety and positive and negative affect. The intervention had no effect on anxiety, affect, or intrusion frequency, but participants in the intervention condition rated their intrusions as less vivid, unwanted, and intrusive. Overall, these findings highlight the need for further exploration of strategies that can protect content moderators against the harmful content they view.

Individuals often search for additional information to make accurate decisions. AI is increasingly valuable as a source of information and advice. Our interest is in AI advice that moderates individuals’ information search. We considered three AI moderators: Devil’s Advocate (DA) encouraged participants to consider counterfactual information; Facilitator (FAC) encouraged consideration of most preferred information; Moderator (MOD) encouraged consideration of least preferred information. Undergraduates played an information search game that required selecting tests, receiving AI advice, and making a decision. Accuracy and test selection strategy were analysed, with and without AI advice. We found that participants adapted search strategies following advice. FAC advice reduced search diversity and duration, while DA and MOD increased both measures. Accuracy was unchanged with any advice. Our study is the first to show that individuals adapt their information search strategy in response to AI advice. Future studies will extend this paradigm to decision-making teams.

To mitigate the negative impact of graphic online imagery, Instagram introduced sensitive-content screens—these screens show blurred images accompanied by a warning. However, a growing body of research suggests these screens do little to deter people viewing, and do not reduce the distress of, harmful content. Here, we examined whether adding emotion regulation instructions to sensitive-content screens improves their efficacy. Participants viewed 30 negative images—each preceded by a sensitive-content screen—and rated their distress for each image. Participants viewed a block of 15 images with no instructions and—after being trained to use reappraisal techniques—15 images with reappraisal instructions, in counterbalanced order. Reappraisal instructions led to lower distress than no instructions, and were more effective when used for the second block of images. Therefore, adding reappraisal instructions to sensitive-content screens can make these screens more effective.

People who use internet search (e.g., Google) to answer questions are more likely to use it again later than people who answer from memory (known as digital fixation). In the current study, we used the Cognitive Reflection Test (CRT) to examine whether digital fixation affects analytical cognitive style. In Experiment 1, participants answered hard trivia questions either by Googling the answers (Google Condition) or attempting to answer from memory (Memory Condition). Participants then switched to answering easy trivia questions and could choose to answer from memory or using Google before completing the CRT. The Google participants chose to answer these easy questions from memory less often compared to the Memory participants, showing digital fixation. Critically, Google participants also had lower scores on the CRT compared to Memory participants. Experiment 2 replicated these patterns, suggesting that using the internet over memory could induce a less analytical cognitive style.