All Subnets

Coldint

#29safety

AI Agent Safety & Security evaluation subnet - benchmarking and scoring AI agents for safety compliance on Bittensor.

Price

$0.014302-0.21%

24h

7d Change

-1.15%

Market Cap

$19.79M

Emission

0.53%

Miners

1

Validators

8

Competitor Mapping

Coldint

Decentralized

Market Cap

$19.79M

TAM

$8.00B

Model

Token-incentivized

Team

Decentralized miners

Anthropic (Constitutional AI) + NIST AI Safety

Private

Market Cap

Private

TAM

$8.00B

Model

Revenue-driven

Team

Enterprise

Coldint aims to be the decentralized counterpart to AI safety evaluation frameworks from Anthropic, NIST, and centralized red-teaming firms, focusing specifically on AI agent safety rather than model alignment.

Implied Valuation

If Coldint captures a share of Anthropic (Constitutional AI) + NIST AI Safety's $8.00B TAM

Bear - 0.1% TAM

$8.00M

implied valuation

-59.58%

vs $19.79M current

Base case

Base - 1% TAM

$80.00M

implied valuation

+304.18%

vs $19.79M current

Bull - 5% TAM

$400.00M

implied valuation

+1920.90%

vs $19.79M current

Scenarios illustrate potential scale relative to the traditional market. Actual outcomes depend on adoption, tokenomics, and competitive dynamics.

Thesis

AI agent safety is an emerging regulatory requirement as autonomous agents proliferate. Being first to build decentralized safety benchmarks on Bittensor could create a valuable niche as governments mandate AI safety testing. However, the pivot from distributed training to safety is still early and unproven.

Team

Led by RWH (PhD in experimental quantum physics) and u (pseudonymous). Both have been in the Bittensor ecosystem since early 2024 with mining experience. Based in Netherlands. Community-driven with Hall of Fame reward system for contributors.

  • RWH - Lead - PhD in experimental quantum physics, Bittensor miner since early 2024
  • u - Co-lead - Pseudonymous, Bittensor ecosystem veteran since early 2024

Funding

No disclosed VC funding. Funded through TAO emissions.

Traction

SN29 token trading at ~$3.12. Validator software live on GitHub with HuggingFace model evaluation. Previous distributed training model struggled post-dTAO launch. Pivoting to AI-ASSeSS (AI Agent Safety & Security) framework.

Recent News

  • 2026 - Relaunching as AI-ASSeSS focused on AI Agent Safety & Security
  • 2026 - Previous distributed training model acknowledged as not working well in dTAO
  • 2025 - Published validator software for model evaluation on HuggingFace datasets

Risk

Team admitted their previous model 'did not work well in a dTAO world.' Pivoting to AI agent safety is speculative with no proven traction yet. Pseudonymous leadership adds uncertainty.