About SussMyAI.com
Understand why SussMyAI exists, how it helps, and who it's built for—all in one place.
AI hallucinations—when models generate incorrect or misleading content—are a known risk. As AI tools become more common in education, research, content creation, and decision-making, the need to verify responses across models has become more important than ever.
Accuracy for AI responses is still a work in progress. Consider the following (sources are listed at the bottom of this page):
- A 2024 survey on Responsible AI engagement showed that 60% of organizations consider inaccuracy to be a relevant concern.
- In a 2023 benchmark test of OpenAI GPT4 it scored in only the 54th percentile for the Graduate Record Examination (GRE) Writing and received a 75% for the Medical Knowledge Self-Assessment Program.
- 2023: A lawyer was fined after citing fake court cases generated by ChatGPT.
We compare your query across four leading AI models (Open AI GPT 4.1 Nano; Anthropic Claude 3.5 Haiku; Meta Llama 4 Maverick (served by Together.ai); and Google Gemini 2.5 Flash). It rates their consistency with a Sus Score:
- High Sus 🛑: Models disagree on key points—watch out!
- Mid Sus ⚠️: At least one model disagrees on a key point—use caution.
- Low Sus ✅: All models are in relative agreement—likely safe.
❗ Always double-check our Sus Scores manually. SussMyAI helps highlight dicrepancies, but accuracy is not guaranteed!
SussMyAI is helpful for:
- Media and Journalists editing content before publishing
- Students, Course Creators, and Researchers verifying information
- Content Generators fact checking AI-generated content
- Developers using AI in applications
- Everyday users unsure if AI is hallucinating
“Sus” is short for “suspicious”—a playful but effective way to label AI answers that may require further investigation. It helps users think twice before trusting AI generated content. "Suss" or "Suss out" means to investigate something so as to gain more knowledge.