The final speaker in this session at the 2026 International Communication Association conference in Cape Town is Ross Dahlke, whose focus is on exposure to untrustworthy Websites in the 2020 and 2024 US presidential elections. According to 2016 data, such exposure is actually fairly limited, but highly concentrated amongst a number of key groups: older adults and political partisans – but (how) has this changed in subsequent elections?
This project captured Web browsing data from a YouGov Pulse panel of some 1,100 participants for four weeks before and one after the respective election dates in 2020 and 2024; this is for both desktop and mobile browsing. These sites were compared against the Lazer Lab and NewsGuard dataset of untrustworthy sites. In 2016, some 44% of people were exposed to at least one such site; in 2020 this dropped to 26% and in 2024 to some 16%. Average visit durations also fell somewhat.
In 2016, older users were especially likely to be exposed to such sites; in 2020 and 2024, this holds, but younger users are comparative more exposed than before. Liberal users were far less exposed in 2020 and 2024, while highly conservative users remained most highly exposed by far. In 2016, Facebook was the greatest source of referrals (at 16%) to such sites; this has largely stopped (3%).
All of these numbers relate to non-AI untrustworthy Websites, though; testing with a new list of AI-generated problematic sites shows that the number of such visits remains low, but increased from 2020 to 2024.
General exposure has continued to go down, then, despite public concerns about untrustworthy sites.
