| What it actually scores | Each verifiable claim in the content — 0-100 credibility score with linked evidence | The account — a bot-likelihood score based on its behavior and metadata |
| Where it works | YouTube, TikTok, Instagram Reels, podcasts, any audio/video link — plus a Chrome/Firefox extension for live watching | X/Twitter accounts only; no video, audio, or transcript analysis |
| Data freshness | Cross-references live web sources at the moment you run the check | Botometer X runs in archival mode on pre-June-2023 data; no scores for accounts created after May 31, 2023 — a consequence of X ending free researcher API access, not a design flaw |
| AI-era coverage | Flags deepfake and AI-generated media signals in the content itself | Per OSoMe's own documentation, the BotometerLite model was trained before modern generative AI and cannot detect AI-supercharged bots |
| Account-level forensics (network, timing, coordination) | None — BSmeter does not analyze accounts, followers, or posting patterns at all | Extracts over a thousand account features across metadata, network, temporal, and content signals |
| Peer-reviewed methodology | No published academic methodology | Peer-reviewed and widely cited; the standard reference tool in disinformation research since BotOrNot launched in 2014 |