AI Security Agents. Continuously find and fix exploits in your code
Joined May 2024
- Tweets319
- Following64
- Followers2.4K
- Likes717
Bitsec | Bittensor Subnet 60 τ retweeted
NEW: @bitsecai has launched Sentios badges: hot links for websites, explorers and dashboards that open its active-projects page. The badges make continuous audit coverage visible where a project is viewed.
teapawt.net/bitsecai/status/211082…
Sentios badges are now LIVE!
Show Sentios coverage anywhere.
Githug, explorers, wallets, dashboards, websites
The badge is now a hot link that opens to our active projects page on the Sentios website:
sentios.io/active-projects
Let the world know Sentios agents are continuously auditing your codebase on significant feature releases.
Want your project listed? Contact us to put Sentios on your codebase.
@bitsecai $TAO
Bitsec | Bittensor Subnet 60 τ retweeted
Can miners find the answers before doing the work?
@yubrew of @bitsecai shows how Bitsec flagged exposed confirmation seeds in @heydittoai's incentive mechanism. He says Ditto fixed the reported issues within days, and explains why seed timing matters.
Bitsec | Bittensor Subnet 60 τ retweeted
Subnet 60 on Bittensor speaking on one of the biggest tech podcasts on earth.
Huuuuge!
$TAO
We will be on TWIST today. Live in 5 minutes @Jason
youtube.com/live/Ynzj1JjoHzE
Bitsec | Bittensor Subnet 60 τ retweeted
The more I see about @ExploitSummit, the more it feels like the best place to experience $TAO beyond the timeline.
The session I’m looking forward to the most is the live security review of other subnets by @Bitsecai.
Their approach to security, including finding ways subnet incentives can be gamed, goes straight at one of the most important questions in #Bittensor: are subnets actually rewarding the right behaviour?
I think more people should be talking about SN60 in general, and I suspect this live demo will get a lot more people paying attention.
Much more interesting than just another slide deck!
We named it Exploit because attacks make networks stronger.
@bitsecai treats security as a permanent adversarial fight. Miners compete to break code. Validators grade on objective truth.
@yubrew is bringing that to the Exploit stage - a live security review of other subnets, in front of the room.
Bring your code.
luma.com/exploitsummit26
Bitsec | Bittensor Subnet 60 τ retweeted
Once-a-year audits are an illusion of safety in a world where attack vectors evolve every day.
Bittensor enables Bitsec to continuously uncover code vulnerabilities. $TAO
Sentios
From the Latin sentire, to perceive, and the Greek -ios, a state of.
A state of awareness. Fitting, for a system that never stops watching.
And it’s more than just our product name. The logo itself is a status badge for your GitHub, your website, signaling your code base is continually audited. Public and visible to anyone, at any time: clients, users, investors.
SENTIOS - ACTIVE
Coming soon, the badge will also link to a live status page showing every project under active coverage with audit dates.
Welcome to Sentios - by Bitsec
@bitsecai $TAO
Bitsec | Bittensor Subnet 60 τ retweeted
A lending protocol is only worth its security.
Mentat Lend already went through two full audit rounds with @burraSec, one of the top auditors in crypto. Every finding is public in our docs, with their fix commits.
Market parameters are frozen at deployment. Logic changes pass through a public 48h timelock.
Now @bitsecai adds continuous audits on top.
Keeping your funds safe is our absolute priority.
Bitsec | Bittensor Subnet 60 τ retweeted
The main one is smart contract risk. Here is how we keep it as low as possible:
Design.
Mentat Lend derives from Morpho, the lending architecture securing billions on Ethereum. Its core contracts have never been exploited.
Audit.
Two audit rounds with @burraSec. Findings and fixes are public in our documentation, all closed before launch.
And now @bitsecai runs continuous audits on Mentat Lend.
Security is our first priority.
Bitsec | Bittensor Subnet 60 τ retweeted
🤖 Made with AI
Here are the auditors who worked on the @MentatLend_ audits 🙏
@burraSec @windhustler
@eeyore0x
@0xT1MOH
@thepantherplus
@adeolRxxxx
Bitsec | Bittensor Subnet 60 τ retweeted
Numerous subnet gems hiding in plain sight:
@engyai
@theminos_ai
@openroboto
@webuildscore
@LeadpoetAI
@bitsecai
@b1m_ai
Great concise SN reports Mark!
Bittensor $TAO is made up of 128 subnets.
The subnets are the meat of the matter. That's where the substance is.
I had my Hermes agent (running Engy kimi k3) write a report highlighting 18 of them.
Bitsec | Bittensor Subnet 60 τ retweeted
Continuous security audit... the way to go! We know it.
The era of single audits is over.
Mission critical systems can't afford to be vulnerable, even for a moment, not when the people attacking are using the best frontier models available.
Continuous, expert-level AI hardening, for every line of code, every release. This is the only strategy that keeps systems safe.
It’s also the premise for everything we built at Bitsec. Priced for continuous auditing of the highest quality, built to secure mission critical systems.
And as Bitsec turns to face the industry with this shift in auditing, we figured it was time to dress for the occasion.
Welcome to Continuous Auditing.
Welcome to Bitsec.
Bitsec | Bittensor Subnet 60 τ retweeted
Right now there's a $110–115 million heist happening to @COLDCARDwallet hw users.
How?
AI found what security researchers missed for 5 years
A bad commit in 2021 in firmware dropped seed entropy to ~40 bits, which generates the seed phrase for wallets. Users thought they were taking the right steps using a high entropy hardware wallet, but really they were unknowingly using weak software PRNG.
Weak PRNG causes collisions every ~1.5 minutes on consumer grade hardware like M1 Mac.
The attacker regenerates seeds offline, and with the newly generated private key, sweeps any funds to wallets they control.
This exploit did not need Mythos or some frontier analysis. It was easily discoverable with even Opus 4.8.
We need continuous auditing to find and fix these issues before hackers do.
Seven weeks ago I pointed Opus 4.8 at the @COLDCARDwallet firmware and told it to find bugs that lose people money.
The second bullet of my prompt: Weak RNG source.
Bitsec | Bittensor Subnet 60 τ retweeted
glad to see rt21 back at it. capricioussage firing up one of the most reliable validators for Bittensor. appreciate your efforts to keep @bitsecai running smoothly
#RoundTable21 lives again (well, kinda!) The OG #RT21 trio you know (and love?) are back with a new $TAO validator called @wildsagelabs 🚀
Help support Real-Infra Network Decentralization on @bittensor (and earn great APY!) by staking to WildSage Labs today!
Bitsec | Bittensor Subnet 60 τ retweeted
$TAO SN60 Bitsec - ELI5.
Nine locksmiths inspect a lock. Two years later, someone arrives with a better tool, opens the exact same lock and steals $5M. The lock did not change. The person looking at it did.
Bitsec has built a Bittensor competition where miners improve AI security agents. The agents are isolated inside sandboxes and tested on Solidity code with critical and high-severity vulnerabilities already confirmed by human auditors.
Finding one bug once is not enough. To pass, an agent must find the complete set in at least two of three runs.
As the agents improve, the same old code can be inspected by a better hunter. The code can stay static. The defense does not have to.
One important distinction. Bitsec did not discover the two incidents cited here. They are showing why an audit should never be treated as the end of security.
Two years. Nine audits. Nobody touched the code.
1inch's Fusion v1 Settlement contract sat untouched since a November 2022 rewrite. In March 2025, someone found an integer underflow in it that nine prior audits never caught, and walked away with $5M. The code hadn't changed once.
Days ago, a different kind of proof. Two OpenAI models broke out of their sandbox through a zero-day in a vendor's background infrastructure. That code hadn't changed either. What changed was whoever was looking.
Most exploited contracts get hit within weeks of deployment. Some sit untouched for years, waiting on nothing but someone capable enough finally looking.
Even an audited codebase that sits still doesn't stay safe. Everything hunting for a way in keeps getting better while it doesn't move.
@bitsecai $TAO
Bitsec | Bittensor Subnet 60 τ retweeted
I looked at bittensor subnets:
#sn96 @verathos_ai (verified inference)
#sn35 @0x_Markets (dex)
#sn60 @bitsecai (agentic code security)
youtube.com/watch?v=M9G_ECZP…
Bitsec | Bittensor Subnet 60 τ retweeted
NEW: Bitsec says SN60 runs top subnet agents against codebases to clear known vulnerability classes, then focuses tokens on edge cases to surface novel exploits.
teapawt.net/bitsecai/status/207457…
ELI5 Bitsec's security approach
We start by running the top agents from our subnet competition on codebases, efficiently clearing out known landmines.
Then, we spend the bulk of time and tokens combing through edge cases to detect new and novel exploits.
In our testing, this approach generalizes well beyond blockchain projects. We find exploits others miss.
Bitsec | Bittensor Subnet 60 τ retweeted
Our IM is the secret weapon that gives us continuously improving results
Bitsec | Bittensor Subnet 60 τ retweeted
Thanks for summing up the @bitsecai opportunity in a short read @taodaily_io 🫡
