Media hype fails to trigger panic over ChatGPT

A new study challenges the assumption that media coverage of emerging technologies inevitably fuels public hysteria. Titled “The next media-fueled moral technology panic? News media’s and audience’s views on ChatGPT” and published in AI & Society (2025), the study explores the rise of ChatGPT in the Dutch media landscape and public consciousness throughout 2023. Drawing on a rich combination of computational content analysis and multi-wave survey data, the research investigates whether ChatGPT prompted a moral technology panic akin to earlier technology backlashes.

Conducted by researchers from the University of Amsterdam, the study analyzes over 2,000 news articles from top Dutch outlets and survey responses from over 1,200 adults across five waves. It probes three key issues: the speed and sources of public awareness, shifts in perceived societal impact, and whether media tone influenced sentiment.

Did ChatGPT awareness follow the pattern of a panic?

Contrary to theories predicting escalating concern and media amplification of technological threats, the study found no substantial evidence of a growing media panic surrounding ChatGPT in the Netherlands. The researchers tracked public awareness, noting a sharp increase shortly after the model’s November 2022 launch. By early 2023, awareness surged to 56% of respondents and climbed to nearly 90% by October 2023.

This rapid increase was attributed, in part, to media exposure. The analysis showed that individuals with greater exposure to ChatGPT-related news were significantly more likely to become aware of it earlier. A logistic regression analysis revealed a substantial gap, nearly 10 percentage points, between high and low exposure groups in the likelihood of learning about ChatGPT. However, while media served as a vector for spreading awareness, this did not translate into alarmist perceptions or widespread concern.

Interestingly, the researchers found that news attention peaked early and then declined moderately over time. This pattern diverges from the standard expectation of steadily increasing media coverage in the early phases of a moral panic. Thus, while media did help disseminate knowledge, it did not sustain or escalate narratives of danger.

Was ChatGPT perceived as a societal threat?

The central tenet of a moral panic involves constructing a technology as a societal threat. To assess this, the study evaluated how both media and the public described ChatGPT’s societal impact. Surprisingly, neither group showed an increasing tendency to portray ChatGPT as powerful or dangerous.

News articles did not present ChatGPT as having a growing influence over time. Analyses of sentiment and linguistic potency, such as use of strong versus weak language, showed no upward trend in framing ChatGPT as a major force. While weak words slightly declined in frequency over time, indicating a shift toward more assertive language, this did not translate into elevated alarm. Public perception also remained relatively stable or even declined in terms of perceived societal impact.

The study further examined whether individuals exposed to media portraying ChatGPT as highly influential were themselves more likely to believe in its impact. No significant effect was found. Mixed-effects models showed that neither sentiment nor strength of language in media coverage predicted respondents’ own beliefs about ChatGPT’s influence on society.

This disconnect suggests that media narratives did not fuel an escalation in public concern. While respondents learned about ChatGPT through the news, they did not absorb a heightened sense of risk or power. Instead, perceptions plateaued or even slightly declined over the 10-month period.

Did negative media tone influence public sentiment?

A hallmark of moral panics is the increasingly negative tone of media coverage. However, the sentiment analysis revealed a different story. Across the board, Dutch news outlets maintained a largely neutral stance toward ChatGPT. In fact, the proportion of negative statements in news articles decreased slightly over time, while neutral statements increased. Positive statements remained consistent.

Public sentiment mirrored this trend. Rather than adopting more negative views, respondents showed a slight increase in positive perceptions of ChatGPT’s societal influence over the course of the year. Regression analysis confirmed that time was associated with a statistically significant, though modest, rise in favorable evaluations.

Crucially, the study found no evidence that exposure to negative or positive news content influenced individuals’ attitudes toward ChatGPT. Even though the media’s tone could be quantified, and individual exposure levels were carefully tracked, no causal link emerged. This undermines a key assumption of moral panic theories - that news framing directly shapes public sentiment.

The researchers suggest that either audiences are becoming more resilient or discerning in processing media narratives, or that moral panic theories may require revision in light of changing media consumption habits and regulatory environments.

Is moral panic a fading framework?

The study’s findings upend several foundational claims of moral panic theory, at least in the context of ChatGPT and the Dutch public. Despite early forecasts from scholars and commentators warning of an impending AI-induced panic, the evidence points to a more measured societal response.

Three key conclusions emerge. First, ChatGPT awareness rose rapidly, highlighting the accelerated adoption cycle of new technologies. Second, no clear signs of panic appeared in either media or public discourse. Sentiment remained balanced, and perceptions of societal impact showed no significant upward shift. Third, the assumed link between media framing and audience evaluation was notably absent.

The authors offer several possible explanations. ChatGPT may simply affect different demographics, such as highly educated professionals, who are less frequently the focus of protective moral narratives. Alternatively, societies and media institutions may have matured in their handling of technological innovations, implementing regulatory discussions earlier and reporting with greater nuance.

Lastly, it is possible that a moral panic has not occurred yet. Historical panics often centered on children and adolescents, and widespread use of ChatGPT among younger populations may not have reached a critical mass. Future research may be necessary to assess long-term developments.