Background: The relationship between media portrayal of psychedelic drugs, scientific research and drug policy is an area of debate.
Aims: To apply artificial intelligence technology to measure trends in media sentiment towards the therapeutic potential of psychedelic drugs.
Method: Up to 300 of the most relevant articles from Google News searches for the term 'psychedelics' were sampled for each year from 2000 to 2025. A large language model, ChatGPT, evaluated subject matter and sentiment.
Results: In total, 88.3% of screened URLs (3308 of 3747) were included in the analysis. The proportion of articles focusing on the therapeutic potential of psychedelics increased from 13.3% (26 of 198) from 2000 to 2009 to 85.3% (1254 of 1470) from 2020 to 2025. The average sentiment score from 2000 to 2025 for articles from all publications (N = 2168) was 78.5 ± 9.3 (mean ± s.d.) (possible range: 1-100). 1.3% (29 of 2168) of articles carried negative sentiment (<50) whereas 4.8% (103 of 2168) had extremely positive sentiment (≥90). Average sentiment reached a peak in 2020 (80.8 ± 7.0), and a statistically significant trough in sentiment was observed in 2024 relative to 2020-2023 (2020-2023, 79.2; 2024, 74.3, P < 0.00001, Mann-Whitney U-test). The proportion of negative-neutral articles (≤65) increased annually from a trough of 3.6% (8 of 267) in 2020 to a peak of 20.9% (43 of 253) in 2024. Artificial intelligence sentiment scores were correlated and concordant with average human rater scores (r = 0.88, concordance correlation coefficient 0.84).
Conclusions: Although most 21st-century media coverage of psychedelic drugs has been positively framed, negative and neutral coverage has increased in frequency since 2020. Researchers, clinicians, regulators and policy-makers should be mindful of the complex relationship between media portrayals of psychedelics and the results of scientific research.
Keywords: Psychiatry and law; philosophy; psychological treatments; psychopharmacology; psychotherapy.