Easily detect LLM generated content using decentralized AI. https://xitter.cf/t.co/cC1Gje8QU7
SN32
Joined March 2024
- Tweets226
- Following36
- Followers1.8K
- Likes298
Pinned Tweet
All you need to know about It's AI (SN32)
We've had a lot of new followers recently, so I decided to put together info about our subnet in one place: what we're building here, key milestones, recent progress, vision and roadmap for the future.
Estimated reading time: ~8-10 min
Our multilingual accuracy results are in 🇦🇪🇩🇪🇫🇷🇪🇸
We benchmarked It’s AI on DetectRL-X (ACL 2026), a new multilingual benchmark covering 6 real-world writing domains, modern LLMs and AI-edited text. ~12K human and AI texts per language:
🇦🇪 Arabic: 99.8%
🇩🇪 German: 99.5%
🇫🇷 French: 99.4%
🇪🇸 Spanish: 96.0%
But accuracy is only half the story. A detector that flags honest work is useless, so we also tested every language on real student writing:
🇦🇪 0.89% false positives (AR-AES)
🇩🇪 0.48% (DARIUS)
🇫🇷 0.56% (E-CALM)
🇪🇸 0.72% (CAES, 20K+ texts)
Under 1% in every language.
In English: 98.3% on RAID and #1 on MGTD, ahead of GPTZero and Originality AI.
Every benchmark, paper and dataset on one page👇
Chrome Extension Upgrade
We've released v2.0.0 version of our chrome extension, which now provide full set of features for AI detection as the main website does.
Alongside with it we've actualised ChatGpt plugin and added support for Claude, so that you could automatically receive AI presence analyses during work on your texts inside both of these platforms.
Check it out: chromewebstore.google.com/de…
It’s AI just got a big redesign.
AI detection isn’t binary, and real-world text rarely is either. So we rebuilt how we analyze and present results:
→ Sentence-level scoring. Every sentence is now classified into one of 5 categories (instead of 3), and the final verdict lands on a 9-level scale: from Human-written to AI-generated with "Mixed", "Likely AI", "Mostly AI" and others in between. More nuance, more transparency, fewer false accusations.
→ Purpose-built interfaces. Each tool now has its own layout designed around what it actually does, instead of one generic results screen.
→ Less friction. Add files to a batch even after a scan has started, restore past scans instantly without re-scanning, and track usage more simply.
This is the foundation for what we’re building next.
We’ve just launched the It’s AI referral program 🎁
Invite a friend to It’s AI:
- They get 20% off
- You earn $7 per referral. Not in credits or extra limits, in USD.
It’s available to every user on the platform. Just log in to your account, and you’ll find your personal invite link there.
Good products grow through word of mouth. Now we’re rewarding it too.
We’ve just rolled out a new lifecycle email engine at It’s AI.
11 automated workflows now cover the full customer journey from activation and conversion to retention and expansion.
The goal is simple: deliver the right message at the right moment to get more payments from users.
It's AI | SN32 retweeted
if you were watching #bittensor subnet channels on discord this week you saw @arbos_born roll out automated ratings for each subnet. presumably these will (eventually) determine arbos weightings for #rootreborn.
these are the FIVE CHEAPEST subnets ranked 10/10 by arbos:
sn55 @NiomeAI - 0.00298 TAO
sn32 @ai_detection - 0.00306 TAO
sn82 @compelleai - 0.00337 TAO
sn18 @zeussubnet - 0.00383 TAO
sn26 @perturbaix - 0.00391 TAO
for context: all of these are 4-5% the price of chutes, which is the highest-priced 10/10 subnet.
full list of 10/10 subnets per arbos is below. bittensor:native
SN32 is a 10/10 subnet. Arbos said it, not me.
World-wide problem, honest competition and a live product with an industry-leading technology behind it.
Multilingual Support at It's AI
We've expanded AI detection to 3 new languages.
French, Spanish, and German are live alongside our existing English and Arabic support. The whole product is translated too.
Already top 5–6 in Google search in the Middle East. Next up: climbing in these new geos.
Onboarding Upgrade
We've updated our onboarding flow and as a result increased Sign Up form completion rate on 15%.
What's done:
- moved to new high-intent login form with specification of why user should sign up and what limit he hitted
- redesigned questionary, removed unnecessary questions/added important ones
- added auto submit on questions and possibility to skip the survey
Keep building📈
Happy to share our second partnership among the bittensor subnets!
It's AI has been integrated into @Bitcast_network to improve the accuracy of their influence mapping and strengthen the platform.
We've run several PR campaigns on BitCast before, so I'm glad we can now help their team as well.
We’ve integrated @ai_detection from Its AI to strengthen the signal and improve the accuracy of influence mapping on @Stitch3_ai.
Pushing the boundaries of creator marketing on X through subnet stacking.
The EU AI Act's transparency rules go live on August 2, 2026. If you build or publish AI content for EU users, Article 50 applies to you regardless of sector or company size.
Four obligations to know:
🔹 Chatbots & interactive AI must tell users they're talking to a machine.
🔹 Providers must mark AI output (text, image, audio, video) in a machine-readable format so it stays detectable as AI-generated.
🔹 Deepfakes must be clearly and visibly labeled.
🔹 AI text published on matters of public interest must be labeled too.
To make compliance easier, the Commission published a voluntary Code of Practice plus an official EU icon set so AI labels look consistent everywhere.
But here's the catch: machine-readable marks only work "as far as technically feasible," and they don't survive copy-paste, paraphrasing, or content from unmarked models. That's where an AI detector like It's AI comes in - helping you spot AI-generated text, decide what needs review, and stay ahead of the rules.
Read full breakdown here: its-ai.org/en/blog/ai-text-l…
Check grammar with It's AI
We're happy to announce that we added grammar checker into our platform!
Since now users can analyse weak spots of the text not only in terms of AI/Plagiarism usage, but also in terms of grammar and see suggestions on how to improve it.
It's AI | SN32 retweeted
AI is getting better at writing daily
The harder question now is, how do we know what’s human?
@ai_detection on Bittensor SN32 $TAO is tackling that problem with a decentralized approach, where models compete to improve AI detection accuracy
Already hitting top benchmark rankings, this is one subnet worth watching as AI content keeps growing
It's AI partnered with BitMind
There are two subnets on Bittensor focused on AI detection: we're leading in AI text detection, while @bitmind focuses on AI photo and video detection. One purpose, different solutions.
Usually, the use cases for these products are different, but when it comes to integrity in general, the combination of both text and image/video detection can be critical.
That's why we've partnered with BitMind. Now, in addition to image/video recognition, they also offer AI text detection powered by It's AI on their platform.
We're excited to announce our partnership with It's AI to advance text deepfake detection capabilities and strengthen our detection systems to include text.
We're working toward a shared goal: making trust and verification more resilient in the age of generative AI.
The future of digital trust depends on verification systems that can evolve as quickly as the threats they are designed to stop. Its AI has the leading AI text detection model and is now available through the BitMind API!
OCR support is live🚀
When standard text extraction fails (scans, photos, broken encodings), we now run visual ML models to pull the text out.
Based on the users' data, this feature will increase the number of processable Arabic documents from 70% to 95%+, since they often have broken encodings. It will also allow you to check PDFs that are actually scans or photos, not just texts.
Alpha Lock Up
Today we locked 600k SN32 alphas (~90% of all owner's alphas) to build the conviction. We genuinely value the belief that you put in us and glad to show by actions our commitment for long-term building as well.
Shaping the future of AI detection together🚀
We've grown x5 in the Arabic traffic for the last month!
Improved our average google position in this segment by almost two spots and gained over 3k clicks from the Google Search to the its-ai.org/ar page.
It's AI is now available at Canvas
We're happy to announce that we've released our Canvas Plugin and now It's AI can be seamlessly integrated into existing educational workflows at schools and universities.
Canvas is a Learning Management System (LMS) that is widely used across the MENA region by schools and universities. All educational processes happen inside this system, such as giving assignments to students, receiving answers from students, grading them, keeping course-related info, etc.
With the It's AI plugin, teachers will be able to automatically see AI scores and deep scan results for students' submissions right inside their LMS — no need to visit our website and do manual checks.
It's a perfect addition to the Educational Enterprise plan, where one solution solves the academic integrity problem without the need to change current grading processes.
