The datasets of Dosimeter are now on Hugging Face

Our three datasets are now on Hugging Face, the open platform where researchers and developers look for AI models and data:
- AI Disinformation Benchmark: 2,783 responses of AI chatbots in eight datasets, with the sources they cited
- AI chatbots and the US midterm elections: 1,000 responses with their sources
- Election-Day AI Poisoning in Germany: 492 responses with every citation and how we classified it
Each dataset on Hugging Face is a copy of our GitHub repository, file for file, so the checksums and our verification scripts give the same result in both places. The data are under the CC BY 4.0 licence, so anyone can use them with a link to the source