ItsAI is a bittensor subnet focused on high-quality AI detection for texts. Recognised as the most accurate AI detector by MGTD benchmark.
Neuron registration trends
The alpha price in τ, day by day, and what moved it each way.
Log change over the last 30 days −7.66%: protocol buys 0.00%, root sells −0.56%, users net −5.07%, cross-subnet −2.03%, unattributed 0.00%.
The chart follows the alpha price in τ. Each period starts at its opening price and steps through the mechanisms below, in a fixed order, to its closing price: a step that rises pushed the price up, one that falls pushed it down, and its height is how far it moved the price. A thin line carries each close into the next period’s open; it is dashed where the two differ. The faint candle behind each period is the market’s own open, high, low and close, for context.
With many periods on screen (90 days, or hours), the steps would be too thin to read, so the chart draws the price as a line instead and colours each period by the mechanism that moved it most. Hover a period for the full step-by-step breakdown.
Not here, because they do not move the spot price: the per-block TAO injection into the pool, and burns.
All figures are log changes, not simple percentage changes, so the parts add up exactly to the total; on big moves the two differ (a 10% log fall is a 9.5% price fall).
2026-11-09 UTC, as of ≈23:54 UTC (block 9,472,061)
The day isn’t over yet, so these figures keep changing. They are in no daily total, average or rank.
Educational institutions, publishers, and content platforms struggle to reliably distinguish human-written from AI-generated text at scale. Existing solutions rely on static models that degrade quickly as language models improve, require centralized infrastructure, and often produce false positives that penalize legitimate human writers.
A distributed network of independently trained detection models, continuously validated against diverse LLM outputs, can maintain accuracy as adversaries improve. By incentivizing miners to develop better classifiers through competitive scoring on unseen data, the system creates natural pressure for innovation without requiring a single organization to maintain and update a monolithic detector.
SN32 implements this by recruiting miners to run detection models (PPL-based or transformer classifiers) against validator-generated challenges. Validators generate synthetic data by pairing human-written texts from the Pile dataset with AI completions from 30+ open-source language models (Llama, Qwen, Mistral, others).
Miners return encrypted predictions scored on F1, false positive rate, and average precision; top performers receive network emissions.
Auto-generated from the GitHub repository and may not reflect the latest changes.
SN32 is a decentralized AI detection subnet that identifies whether text was human-written or AI-generated. It provides binary classification and sentence-level probability scoring. The subnet operates with miners running inference models to classify text and validators generating synthetic data from 30+ LLMs to evaluate miner performance.
According to MGTD benchmark (ICAIE 2025), it achieves 92% ROC-AUC accuracy, ranking first among AI detectors.
The subnet uses a validator-miner architecture. Validators (using Ollama to run 30+ diverse LLMs) generate synthetic text challenges mixing human-written texts from Pile dataset with AI completions. They send encrypted batches of text to miners via TextSynapse protocol.
Miners load pretrained models (PPLModel based on Phi-2 with linear calibration, or DebertaClassifier) and return encrypted per-token or per-word predictions. Validators decrypt, score miners on F1/FP/AP metrics, and update weights. Miners are blacklisted if hotkey is unrecognized or stake falls below threshold; priority is determined by caller's stake in metagraph.
Validators generate balanced 1:1 human/AI text pairs with diverse LLMs (Llama, Qwen, Command-R, Mistral, etc.) and random generation parameters. Human texts come from Pile dataset; AI texts are generated completions. Augmentations include misspellings and random adjective removal to prevent memorization.
Miners' predictions are encrypted to validators' ephemeral keys. Three scoring dimensions: (1) F1-score for overall classification accuracy, (2) False Positive rate (1 - FP/samples) weighted higher to avoid wrongly penalizing humans, (3) AP score from precision-recall curve. Out-of-domain validation uses exponential moving average with alpha=0.2.
The sections above were auto-generated from the GitHub repository and may not reflect the latest changes. View repository (github.com, opens in a new tab, external site)
ItsAI is a bittensor subnet focused on high-quality AI detection for texts. Recognised as the most accurate AI detector by MGTD benchmark.
ItsAI is a bittensor subnet focused on high-quality AI detection for texts. Recognised as the most accurate AI detector by MGTD benchmark.
This information was auto-extracted from the GitHub repository and may be incomplete or out of date. Always refer to the subnet's GitHub repository (github.com, opens in a new tab, external site) for the latest instructions.
Sources disagree: GPU model, VRAM
Not yet verified: Notes
Not yet verified: Notes
Machine-extracted from the subnet's repository docs; may be wrong or out of date. Check the repository before buying or renting hardware. Open the repository (github.com, opens in a new tab, external site) · Extracted 2 Nov 2026
Clone repository and update system packages
apt update && apt upgrade -y && git clone https://github.com/It-s-AI/llm-detection
Create and activate Python virtual environment (Python 3.8+)
python -m venv venv && source venv/bin/activate
Install detection package and dependencies
cd llm-detection && rm -rf detection.egg-info && python -m pip install -e .
Verify correct packages installed
python -c "import bittensor, detection; print('ok', bittensor.__version__)"Download pretrained model weights
wget https://huggingface.co/sergak0/sn32/resolve/main/deberta-large-ls03-ctx1024.pth -O models/deberta-large-ls03-ctx1024.pth && wget https://huggingface.co/sergak0/sn32/resolve/main/deberta-v3-large-hf-weights.zip -O models/deberta-v3-large-hf-weights.zip && unzip models/deberta-v3-large-hf-weights.zip -d models/deberta-v3-large-hf-weights
Create Bittensor wallet and register hotkey on subnet 32
btcli w new_coldkey && btcli w new_hotkey && btcli s register --netuid 32 --wallet.name YOUR_COLDKEY --wallet.hotkey YOUR_HOTKEY
Install PM2 process manager
sudo apt update && sudo apt install jq npm && sudo npm install pm2 -g && pm2 update
Start miner with PM2
pm2 start --name net32-miner --interpreter python3 ./neurons/miner.py -- --wallet.name YOUR_COLDKEY --wallet.hotkey YOUR_HOTKEY --neuron.device cuda:0 --axon.port 70000
Optional: Run local Subtensor node for better performance(optional)
git clone https://github.com/opentensor/subtensor.git && cd subtensor && docker compose up --detach
Clone repository and update system packages
apt update && apt upgrade -y && git clone https://github.com/It-s-AI/llm-detection
Create and activate Python virtual environment
python -m venv venv && source venv/bin/activate
Install detection package with custom mathgenerator fork
cd llm-detection && python3 -m pip install -e . && python3 -m pip uninstall mathgenerator -y && python3 -m pip install git+https://github.com/synapse-alpha/mathgenerator.git
Create Bittensor wallet and register hotkey on subnet 32
btcli w new_coldkey && btcli w new_hotkey && btcli s register --netuid 32 --wallet.name YOUR_COLDKEY --wallet.hotkey YOUR_HOTKEY
Export Weights & Biases API key for logging
export WANDB_API_KEY="<YOUR_WANDB_API_KEY>"
Install PM2, lshw, and build dependencies
sudo apt update && sudo apt install jq npm lshw build-essential cmake libboost-system-dev libboost-thread-dev libboost-program-options-dev libboost-test-dev zlib1g-dev libbz2-dev liblzma-dev zip unzip -y && sudo npm install pm2 -g && pm2 update
Install Ollama and start service in background
curl -fsSL https://ollama.com/install.sh | sh && pm2 start --name ollama "ollama serve"
Build and install cc_net for data processing
cd cc_net && export TAR_OPTIONS="--no-same-owner" && pip install -e . && make install && make lang=en dl_lm && cd ..
Start validator with PM2
chmod +x run.sh && pm2 start run.sh --name llm_detection_validators_autoupdate -- --wallet.name YOUR_COLDKEY --wallet.hotkey YOUR_HOTKEY --axon.port 70000 --neuron.device cuda:0
| Day (UTC) | Share |
|---|---|
| 2026-11-08 | 0.01% |
| 2026-11-07 | 0.01% |
| 2026-11-06 | 0.01% |
| 2026-11-05 | 0.01% |
| 2026-11-04 | 0.01% |
| 2026-11-03 | 0.01% |
| 2026-11-02 | 0.01% |
| 2026-11-01 | 0.01% |
| 2026-10-31 | 0.01% |
| 2026-10-30 | 0.01% |
| 2026-10-29 | 0.01% |
| 2026-10-28 | 0.0047% |
| 2026-10-27 | 0.01% |
| 2026-10-26 | 0.01% |
| 2026-10-25 | 0.01% |
| 2026-10-24 | 0.01% |
| 2026-10-23 | 0.01% |
| 2026-10-22 | 0.01% |
| 2026-10-21 | 0.01% |
| 2026-10-20 | 0.01% |
| 2026-10-19 | 0.01% |
| 2026-10-18 | 0.01% |
| 2026-10-17 | 0.01% |
| 2026-10-16 | 0.01% |
| 2026-10-15 | 0.01% |
| 2026-10-14 | 0.01% |
| 2026-10-13 | 0.01% |
| 2026-10-12 | 0.01% |
| 2026-10-11 | 0.01% |
| 2026-10-10 | 0.01% |
| 2026-10-09 | 0.01% |
| 2026-10-08 | 0.01% |
| 2026-10-07 | 0.01% |
| 2026-10-06 | 0.01% |
| 2026-10-05 | 0.01% |
| 2026-10-04 | 0.01% |
| 2026-10-03 | 0.01% |
| 2026-10-02 | 0.02% |
| 2026-10-01 | 0.01% |
| 2026-09-30 | 0.02% |
| 2026-09-29 | 0.01% |
| 2026-09-28 | 0.01% |
| 2026-09-27 | 0.01% |
| 2026-09-26 | 0.02% |
| 2026-09-25 | 0.01% |
| 2026-09-24 | 0.01% |
| 2026-09-23 | 0.01% |
| 2026-09-22 | 0.01% |
| 2026-09-21 | 0.01% |
| 2026-09-20 | 0.02% |
| 2026-09-19 | 0.01% |
| 2026-09-18 | 0.01% |
| 2026-09-17 | 0.02% |
| 2026-09-16 | 0.02% |
| 2026-09-15 | 0.02% |
| 2026-09-14 | 0.02% |
| 2026-09-13 | 0.02% |
| 2026-09-12 | 0.02% |
| 2026-09-11 | 0.02% |
| 2026-09-10 | 0.02% |
| 2026-09-09 | 0.02% |
| 2026-09-08 | 0.02% |
| 2026-09-07 | 0.02% |
| 2026-09-06 | 0.02% |
| 2026-09-05 | 0.03% |
| 2026-09-04 | 0.03% |
| 2026-09-03 | 0.03% |
| 2026-09-02 | 0.03% |
| 2026-09-01 | 0.02% |
| 2026-08-31 | 0.02% |
| 2026-08-30 | 0.02% |
| 2026-08-29 | 0.02% |
| 2026-08-28 | 0.02% |
| 2026-08-27 | 0.02% |
| 2026-08-26 | 0.02% |
| 2026-08-25 | 0.03% |
| 2026-08-24 | 0.03% |
| 2026-08-23 | 0.02% |
| 2026-08-22 | 0.03% |
| 2026-08-21 | 0.03% |
| 2026-08-20 | 0.03% |
| 2026-08-19 | 0.04% |
| 2026-08-18 | 0.03% |
| 2026-08-17 | 0.03% |
| 2026-08-16 | 0.03% |
| 2026-08-15 | 0.03% |
| 2026-08-14 | 0.03% |
| 2026-08-13 | 0.03% |
| 2026-08-12 | 0.03% |
| 2026-08-11 | 0.04% |