Bitsec | Bittensor Subnet 60 τ

AI Security Agents. Find and fix *all* exploits in your code

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

https://t.co/YBJkTxaepH

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

We just got our biggest batch of security agents, 98 up from 70 in the past round.

It's very competitive between 5-6 miner groups.

Final results in the next 24 hours.

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

Security is the process of sweeping your code for every potential landmine. Miss one, your protocol might explode.

We found stacking different agent findings together increases coverage. It's something important that many people are missing.

1. We can point our IM towards undetected exploits, and add those new agents to the stack.

2. We can improve cost performance on common exploits to agents run on smaller, faster models.

3. Currently, our security swarm of agents running on production code is finding criticals and highs in pretty much every run.

We already find security exploits today. With our competitive 5-6 miner groups, the swarm gets stronger every week.

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

https://t.co/iaucvb6zgM

[TensorUSD (TUSDT) @TensorUSD]

Bittensor validators contribute to price feed quality on SN113.

But before any price touches a TensorUSD vault, it goes through the OSM first. Buffered. Validated. If something looks wrong, emergency oracles step in. Governance can pause the feed entirely.

The intelligence layer and the oracle layer work together. That's what Bittensor-native actually means.

Do you think native oracle validation is better than third party price feeds? 👇

[yubrew @yubrew]

✅ locked

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

640332.69 Bitsec tokens are locked by the team.

Over $600,000. Perpetually locked.

Remember that mining activity results in better agents, results in more revenue to increase token value.

Incentives are aligned.

[SubConnect @WeAreSubConnect]

Miner changes for Bitsec that have a direct effect in results

[taostats τ @taostats]

NEWS: SN60 @bitsecai achieved a new milestone with stronger agent performance and faster vulnerability detection after the v3.1 update.

Top-performing agents found every critical vulnerability in over half of tested projects.

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

@vitobotta @taostats It becomes a game of evals and data science. A suite of narrow scoped agents + verification engine progresses towards 100% coverage. Business logic flaws are easy for agent based detection

[Yuma @YumaGroup]

A lot happened on Yuma-accelerated subnets in May. some highlights:

1️⃣ SN23 @trishoolai for AI security landed SN64 @chutes_ai as a client, and had parent company Astroware accepted into the @nvidia Inception program.

2️⃣ SN44 @webuildscore for video interpretation released a scaled scoring model with a cricket delivery challenge.

3️⃣ SN59 @babelbit for translation benchmarked at 80% faster than Google, and rolled out a new look.

4️⃣ SN60 @bitsecai for agentic code review shared compelling open-source competitive results.

5️⃣ SN61 @_redteam_ for device-level identification leveraged to investigate $1.5B Bybit incident.

6️⃣ SN105 @b1m_ai for bandwidth coordination launched Transfer Studio for institutions, and went live with Prism for better aligned rewards scoring.

If you're building with AI, give them all a look. More exciting news on deck this week at the @proofoftalk Bittensor track.

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

We dug into a frustrating problem: more than half of the agents in a recent round scored a flat zero. Not because the agents were bad, but because of small, avoidable config errors.

A bad API key here, an unreliable inference provider there, a model chosen right at the threshold that tanked under load.

Miners were putting in real work and walking away with nothing to show for it, often without knowing why.

We traced it back to the root issue. The common thread was that these failures only surfaced during evaluation, long after there was any chance to fix them.

Our fix moves the catch earlier. We added a live project to the screening step that runs the agent end to end, confirming it actually executes, logs properly, and produces real output. The screener is the right home for this because it runs during the submission phase on our validator, which means miners get instant feedback while they still have time to correct course and resubmit.

The result:

Miners catch problems before they cost anything, they stop burning money on doomed evaluation runs, and the subnet progresses much faster.

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

We introduced a bunch of miner friendly changes this round, it's directly lead to more miner participation.

47+ agents this past round with the top agent scoring over 50%.

The agent is able to find ALL critical and high vulnerabilities in over 50% of the projects on open source models, in less than 30 minute run time.

We are adding brand new challenge problems for the next round and tuning the IM so future rounds even more useful to revenue generating customers.

[Swamination @Swamination]

Always a great time with @VenturaLabs, talking all things #Bittensor! https://t.co/0rfxZYKPSf
Shoutout on the security segment: a few teams working to solve security challenges:
@_redteam_ - Leveraging intelligence for secure decision-making
@bitsecai - Unearthing hidden threats within code
@trishoolai - Monitoring AI alignment (Halo Guard now in @Chutes!)
@yanez__ai - Protecting the human in an AI world

@YumaGroup

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

Most votes confirm bitsec is legit.

Looking through the downvotes, there is at least 1 freelancer who was not hired, the rest have never posted or DM'ed the subnet team.

We're open to criticism and feedback, but strange we're getting ghost downvotes.

Fair?

[yubrew @yubrew]

There are many “sleeping beauty” subnets burning 100% emissions, owner self mining to move miner emissions to owner wallets, gaming tap flow and other predatory actions.

only took 1 msg to wake them up

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

AI security is a big area. We haven't spoken to trishool yet but I think we are handling different areas.

Bitsec is working on cybersecurity. Finding and fixing security issues, using a flexible IM to improve coverage, proof of work.

[biττenSOAR @bittenSOAR]

@gordonfrayne @trishoolai 1. Discuss their parent company @astrowareai

2. Most of what they discuss recently is AI security. How do they differ from @bitsecai?

3. Where does AI alignment fit in? Ensuring AI is doing what we want. Aligning with our values.

4. How do miners contribute to alignment?

[YVR τrader @YVR_Trader]

Great progress over on @bitsecai #SN60

[yubrew @yubrew]

current view of chutes vs openrouter for our miners. unfortunate but true.

you're putting yourself at a big disadvantage using chutes unless you really, really know the platform well.

[yubrew @yubrew]

would it be interesting if we did more in depth post round analysis?

btw download the top agent, see trace logs and run it yourself https://t.co/VdIav2fqjm

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

The top agent scored 38.1%, over 10%+ bump in the bitsec leaderboard since the last round. Miners are doing an increasingly better job at writing better agents.

We got a few things done last week to make miners' jobs easier. Mostly UX changes. Now just visit the site to see when rounds start / stop, how much emissions current winner gets, countdown.

We manually tested several rounds and have the process down. This week we automated everything.

All agents go through screeners before evaluation, allowing miners instant feedback. Submission time is scheduled with automatic start / end times. Emissions kick off automatically after all evaluations are done.

All agents are public after evaluations are done to kickstart the next round of research and competition. Now we also have downloadable trace logs and output logs to help boost agents even faster. We also gave a much needed update to the docs.

Next week we're working on heavier infrastructure changes to make the IM more flexible.

Onwards and upwards!

[Bitsec | Bittensor Subnet 60 τ @bitsecai]

Current Round evaluations have started. Miner agent submission window is closed.

There's an uptick in agent activity, 46 agents submitted for round v3.0.2

Historically takes 1-2 days to finish evals. How do you think the miners did?