Bittensor Subnet SN61 RedTeam Boosts Bot Detection Accuracy to 99.5%

Bittensor (Bittensor) Subnet 61 (SN61) 'RedTeam' has raised bot detection accuracy from 74.3% to 99.5%. The platform, operated by cybersecurity firm Innerworks, achieved this through 11 months of continuous adversarial testing. Crypto Briefing reported on July 7 (local time) that Innerworks announced these results.
RedTeam, launched at the end of 2024, is a decentralized cybersecurity challenge platform where miners submit Docker-based security solutions for evaluation in controlled environments, earning TAO tokens as rewards based on performance. The platform has run 14 challenges across various scenarios including human behavior mimicry and automation bypasses, with nine challenges exceeding difficulty thresholds and being replaced by new ones.
The 99.5% figure represents cumulative results from the 11-month testing period. Innerworks noted that competing systems' average performance dropped to 14.6% during the same period, though they did not disclose comparison methods, timeframes, or metrics. The platform has integrated with DeFi aggregator protocol 1inch and messaging applications with over 100 million daily active users, while five enterprise applications with combined market capitalization exceeding $22 billion (approximately KRW 29.634 trillion) and processing over one billion transactions per day are in testing.
Bittensor is a decentralized machine learning network capable of infinitely expanding domain-specific subnets. Each subnet operates independently with its own miner-verification structure and token incentives, with RedTeam being cited as an early example applying this structure to cybersecurity. Unlike traditional proof-of-work (PoW) mining that allocates computing resources for block creation, miners here contribute to solving security challenges in exchange for TAO tokens.
The approach of operating bot detection as a competitive task within decentralized networks differs from centralized security vendors developing solutions internally. However, Innerworks noted that further validation is needed to confirm whether results from controlled testing environments will translate to real-world production settings.
Innerworks announced plans to develop the 'Immune System' framework, expanding beyond detection-focused approaches to real-time attack-defense operations. Key components include the 'Bot Virus' challenge, which aims to have miners develop adversarial agents capable of simultaneously conducting attacks and defenses. Additional features include encrypted submission methods to prevent code replication among miners and enhanced incentive structures for sustained contributions.
Regarding economic models, Innerworks mentioned a token buyback plan funded by SaaS revenue, lockup profits, and future token emissions. However, no specific release dates or buyback scale were disclosed for the Immune System framework or Bot Virus challenges. Community reactions from Bittensor or cybersecurity industry responses to this announcement remain unreported as of now.
The 99.5% accuracy rate was achieved in controlled experimental environments, and performance in real production settings or live attack scenarios has not been verified. Integration cases with partners like 1inch are currently limited to testing phases, with the transition of five enterprise applications into actual use remaining a future milestone for verification.
Korean Source
This article is an English localization of a Korean-language crypto news report. Original headline: 봇 탐지율 74%→99.5%…비트텐서 채굴자, '화이트해커'로 바뀌었다