Uploads images and capscreens to check weather the information is legitimate or not. Its help detect deepfake images and scam patterns.
We are still explore how we can make the product better! We need your comments and feedbacks on how we can defeat Scammers via AI.
We aim to build the world largest anti-scam platform and community for people and businesses.
@seungwhan We are actively collecting data for training semantic analysis for scam. For now, to be honest , we can't cover all cases, because different regions have different scam patterns, but we have confident to tackle them.
@dennis_ng1 congrats! This is interesting :) Would you be internet in making your API available through our Marketplace? I think it would be a good addition to the existing catalog. We do have other related APIs but not a comprehensive anti spam one. Would be keen to explore.
@sonu_goswami2 Thanks for supporting, we do need everyone to suggest and giving us feedback, then we can discuss with the team and community that how we can proceed into next milestone.
@ashish_parmar13 Hi Ashish, we do want to build real-time scam alerts and integrate with other platform. Because of the model requires 5-10s to verify images, so we need to improve latency to archive real-time detection. We are build plugin for chrome for easier user experience, this Beta version PH launches that we want feedback from to community to solidify the future development path.
Report
With scams moving rapidly toward deepfake docs, images, video and voice, how are you prioritizing where to invest dev effort? Are you weighting detection by prevalence today vs. what you expect to spike next? Curious how you're thinking about sequencing.
I like the idea behind this. The question is: What is the accuracy of the estimates of such a tool? How does it work to be accurate in identifying some threat?
@busmark_w_nika Hi Nika, thanks for supporting , we are in exploring stage for this idea too. Basically, we wanna use AI methods to detect scams including text and images, thats why we have the beta launch to get more insight and feedbacks for this idea. For the accuracy, I will ask my teammate to answer this part with more statistic details later.
Thats an awesome tool! When you mentioned "verify messages" how do u achieve that? With NLP or something else? can it be used directly on telegram for example? where the majority of the scammers live
@cryptosymposium Thanks for support!
We use multiple layers to verify messages, including NLP and LLM. For now we can only use capscreen to verify on the platform. We understand that normally user will not verify their info in third party website in to c. So, we are also thinking to build telegram or whatsapp chatbot to help people resolve this scamming information problem.
Report
Interesting idea. Are you also thinking about adding encryption features (giving a certificate of sorts) as a token of proof or something?
@utkarsh_upadhyay4 yes we will get Soc type 2 and GDRP for cert as soon as possible
Report
Congrats @dennis_ng1! Having navigated the murky waters of online misinformation, I find your approach of verifying images and screenshots incredibly timely. One question though, how do you plan to handle the inevitable false positives? I'm in for beta testing and feedback!
Report
@david_xu7 Thanks for the insight, David! It’s great to see that you’re prioritizing that in your model optimization. I’m excited to see how the model evolves in this area
Report
Interesting launch! Curious if we can get an idea of how threats are caught or identified and what that workflow looks like - a little more detail in the description would be helpful!
@dzaitzow for images that contains faces, we trained an model specifically targeting deepfake. The rest text will be extracted by ocr then run through nlp
Scam AI
Ollie
Scam AI
ApyHub : The All in one API Platform
Scam AI
Scam AI
With scams moving rapidly toward deepfake docs, images, video and voice, how are you prioritizing where to invest dev effort? Are you weighting detection by prevalence today vs. what you expect to spike next? Curious how you're thinking about sequencing.
minimalist phone: creating folders
Scam AI
chatWise
Scam AI
Scam AI
Scam AI