AI-driven deep security intelligence for complex systems.
Joined September 2025
- Tweets83
- Following5
- Followers1.4K
- Likes191
Close to 1,000 people registered for Risky Business, Singapore. Only 110 made it into the room.
Tight curation makes for great conversations, and plenty of great conversations were had.
Thank you to the builders, protocols and capital allocators who kicked off their @token2049 week with us.
See you at the next one. 🫡
GregoAI retweeted
A note to those building on chain in 2026:
vibe coding + high code velocity + bull market pace + rapid TVL growth = bugs galore
we see many teams building skyscrapers on sand with a "fix it later" mentality
initially this works and they think they are geniuses
... but eventually
someone will find the big, protocol wrecking, critical bug
capital flight and brand destruction happens overnight
entire savings lost, lives ruined financially, lawsuits filed, death threats received
i've seen it happen time and time again and it will happen to YOU if this is your approach to building on the blockchain
riptide's recommendations:
1) find a balance between securing the code and shipping fast. the security strategy should not be to push critical fixes like OTA updates for an XBOX game when any significant TVL is at risk.
2) invest in top tier human security talent. AI + expert SRs are a force multiplier. get a team that builds with a security first mindset from day 1.
3) properly incentivize responsible disclosure and treat your whitehats like kings. they are your last defense and will go out of their way to ensure your success if treated well
remember that you may only get one shot at building a successful project that makes you millions of dollars ...
best of luck anon
GregoAI retweeted
🚨We now have a live website for new guests that would like to come on the podcast! 🚨
You can self-submit and I will review all submissions personally. The idea is to build a queue of potential guests so that I can put out more episodes on a consistent basis going forward.
Check it out here: bountyhunt3rz.fm/
Thank you to the champion @khubak for putting this whole thing together
Alpha confirmed
if you will be at @token2049 in SG this year
our team @therealgregoAI will be hosting an event here: luma.com/dupkf78j
GregoAI retweeted
New @cyphertalkmedia episode: our co-founder @beyer_st sits down with @0xriptide of @therealgregoAI.
Chaining vulnerabilities with AI, the $27.7M exploit it caught, and what's left for the human auditor.
🎙️ New episode: @0xriptide, bug bounty hunter turned co-founder and CEO of Grego AI.
He built AI that chains vulnerabilities humans would almost never reach, found a high-severity bug in Lido, and prevented a $27.7M exploit.
🎧 podbean.com/ep/pb-e9u3w-1b5f…
GregoAI retweeted
If you think AI security tools are just cheap scanners that dump 20 pages of false positives, this conversation is worth listening to.
This week’s Security Voices features @flack00n, Head of Bugs at @therealgregoAI .
From @bountyhunt3rz with @0xriptide.
What I found interesting wasn’t just the $500K+ in bounties or the $27.7M they helped save.
It was how they think about depth, verification, and why they still don’t trust AI 100%.
Some thoughts from the conversation 👇
Finding bugs that humans miss ... while staying extremely humble
An AI security agent now ranks #17 on the Immunefi leaderboard.
In 2026 alone, it has found 3 valid Critical vulnerabilities and 5 Highs, earning more than $197,000 in bounties.
The AI security agent race is clearly not slowing down.
If you’re building an AI security agent and have been hunting on Immunefi, DM us. We’d love to see what you’re working on, support your training with our Studio Review product, and help amplify your work.
Congrats, @therealgregoAI! 🤖
GregoAI retweeted
An AI security agent now ranks #17 on the Immunefi leaderboard.
In 2026 alone, it has found 3 valid Critical vulnerabilities and 5 Highs, earning more than $197,000 in bounties.
The AI security agent race is clearly not slowing down.
If you’re building an AI security agent and have been hunting on Immunefi, DM us. We’d love to see what you’re working on, support your training with our Studio Review product, and help amplify your work.
Congrats, @therealgregoAI! 🤖
GregoAI retweeted
Shocked seeing incentives misaligned yet again
$320M or potential $10k w negotiation + KYC
No, Liquid Network / Blockstream did not run a public security bug bounty program.
Elements (the software powering Liquid) and related projects only had a standard responsible disclosure policy: report privately via PGP-encrypted email to securityblockstream.com. No rewards or formal bounty platform (Immunefi, HackerOne, etc.) were offered for vulnerabilities.
They occasionally posted development bounties for features (e.g. wallet kits), but nothing covering protocol or consensus bugs.
Join us in Amsterdam during @CryptoCanal and discuss bugs over beers w/ @0xriptide
luma.com/jafnlbbu
GregoAI retweeted
An AI just found a critical bug worth $70,000.
That brings its 2026 earnings past $187k.
Pledge $IMU behind @therealgregoAI and you both earn when it finds the next one: immunefi.com/pledge/gregoai/
GregoAI retweeted
The real bottleneck in AI security isn't prompting. It's reasoning.
@0xriptide, CEO of @therealgregoAI and a top-40 globally ranked bug bounty hunter, joins the Common S3nse stage.
After years of finding high-severity vulnerabilities in smart contracts and DeFi, Justus is now building AI-powered defenses that can reason, verify, and find real bugs without drowning teams in false positives.
Privacy • Security • DeFi
📍 Amsterdam
📅 September 4-5 (during Cypherpunk Week)
🦦 Privacy is Common S3nse 👉 commons3nse.cryptocanal.org/
🐈⬛ Cypherpunk Week 👉 cypherpunkweek.com/
GregoAI retweeted
BOUNTYHUNT3RZ Episode 36: w/
@flack00n
&
@0xriptide
discuss the inner workings of @therealgregoAI's lab including the origin story, why riptide co-founded the company, the AI security narrative, the future of bug hunting with AI, how their depth-first tool stacks up against the rest of the AI-sec space, flacko's humble beginnings and his journey with Grego AI, random discussion about motorcycles, and much, much more …
podcast, discord, substack: linktr.ee/bountyhunt3rz
Glad to receive some humble donations from
The DAO Fund's First Security Round on
@Giveth
We have put this funds to good use in creating the next iteration of Grego AI by funding continued R&D and covering a portion of our inference costs
Thank you also to the organizations who contributed to the funding including @wintermute_t and @thedaofund
GregoAI retweeted
We will see many more of these on new chains and low activity chains
Ethereum and its L2s have been tested out in the open, thus AI blackhats have moved on to any low hanging fruit that has never really had eyeballs but secures value
Yesterday was the time to think about and implement a strong AI native defense security strategy, unfortunately most will wait until it’s too late
Wanchain @wanchain_org Cardano bridge was reportedly being attacked, with ~515M $NIGHT drained from the bridge Treasury.
Our initial investigation suggests that the root cause seems to be a non-injective signed-message encoding in the TreasuryCheck validator. The signed message is built by raw-concatenating 14 variable-length redeemer fields (via `AppendByteString` fold) without delimiters or length prefixes. Different field-value tuples can produce the same byte string — and therefore the same hash and a valid signature reuse.
We verified this by decompiling the on-chain Plutus V2 bytecode (pic 1) and decoding the attack redeemer from the exploit TX (pic 2).
The same on-chain bytecode also contains a `SerialiseData` builtin — but it is only used on the output-datum matching path, not in the signature-hash construction. Using `Sha3_256(SerialiseData(...))` instead would have provided unambiguous CBOR-encoded field boundaries, preventing this class of field-splitting signature reuse.
Attack TX: cardanoscan.io/transaction/0…
