Now Live on X
Meet @BaloneyAI
Mention the bot on any post. Get an instant AI detection verdict. Text or image -- we scan it, break down exactly how we got there, and never charge you a dime.
Follow @BaloneyAIHow it works
Reply or quote-tweet with @BaloneyAI
Tag the bot on any post you want checked. Works on text posts, image posts, and quote tweets.
We run 20 detection methods
Watermark checks (C2PA, SynthID), ML classifiers (Oculus 2.0, TMR), commercial APIs, forensic analysis, and statistical signals -- all in parallel.
You get a public verdict
A reply with the verdict, confidence score, and the top methods that contributed. Transparent by default -- anyone can see how we reached the conclusion.
What the replies look like
AI GENERATEDText scan complete
Confidence: 99%
Pangram: 99% | Oculus 2.0: 97% | TMR: 98%
baloney.app/methodology
LIKELY HUMANImage scan complete
Confidence: 92%
C2PA: Camera verified | EXIF: Canon R5 | Noise: Plausible
baloney.app/methodology
UNCERTAINImage scan complete
Confidence: 54%
Hive: 48% | Frequency: 61% | Metadata: No EXIF
baloney.app/methodology
AI GENERATEDImage scan complete
Confidence: 95%
SightEngine: 96% | ELA: Suspicious | C2PA: AI origin declared
baloney.app/methodology
Every reply links back to our methodology page so you can see exactly how detection works. No black boxes.
Not your average AI detector
20 methods, not one
Watermarks, ML classifiers, forensic analysis, statistical signals. If one method misses it, another catches it.
We verify humans too
Our Human Provenance Score checks camera signatures, sensor noise, lens data, and GPS. Positive proof, not just absence of AI.
Transparent verdicts
Every result shows which methods fired and their individual scores. We'd rather say 'uncertain' than get it wrong.
Free forever on X
The bot is free. No signup, no paywall, no catch. Mention @BaloneyAI and get a verdict. That's it.
Built different. On purpose.
Baloney is built by BNVG -- the same team behind Pavlo, a cryptographic agent attestation platform. Every detection result can be signed with an Ed25519 receipt: tamper-proof, independently verifiable, legally admissible. We don’t just tell you something is AI -- we prove it.
Ready to call it out?
The bot scans posts. The extension scans your entire feed. Use both.