Pinned Tweet
Gm CT ☀️
Happy Saturday, ya all touching some grass today?
Bolex 🀄 retweeted
Good morning brother and sister
Been grinding content around early projects for a while and @rarelistio kept showing up differently than the usual campaign noise.
You pick a vetted project, post about it on X, finish the tasks, and the system tracks your contribution into RARS points that move you up the leaderboard.
Season 1 has been running since early October with a $50,000 prize pool and over ten thousand creators already in. RARS are set to convert into the project’s own $RARE later.
What stood out after watching the recent campaigns is how quickly real rewards moved. Project Sphere finished and the top twenty each received 100 USDT straight to the wallets they submitted.
The top two hundred are locked for a share of the token supply, and the team already signaled another campaign right before launch. That part felt more concrete than most point systems I have seen.
They also spent the last stretch reworking how content gets scored. The next wave of campaigns will use the upgraded algorithm that prioritizes quality over pure volume or repetitive posts.
The stated goal is simple: give more people a realistic chance to earn instead of only the biggest accounts. Filtering low-effort content has been tightened at the same time.
I have been testing it by writing about the live campaigns rather than just tagging and hoping. The tracking is automatic once you connect, and the leaderboard updates cleanly enough to see whether a post actually moved the needle.
Competition is tight near the top, but the structure rewards consistency and useful posts more than spam. Smaller accounts can still climb if the content lands.
The conversion details for RARS into $RARE are still open, and the exact season end date remains TBA, so I treat the points as upside rather than a guaranteed claim. The USDT payouts that already went out make the short-term side more interesting while waiting.
If you create content regularly and want to contribute to projects that actually get reviewed before they go live, this setup is worth testing for a week or two. The username is @rarelistio
🤝 Paid partnership
Các quỹ chính trị ủng hộ crypto đã chi hơn 63 triệu USD để chống lại Sherrod Brown ở Ohio. Ông vẫn thắng.
Trong bầu cử giữa kỳ Mỹ, phe Dân chủ giành Hạ viện. Fairshake cũng chi hơn 10 triệu USD chống Juliana Stratton ở Illinois, bà cũng thắng. Quỹ liên quan đến Tether và Cantor ủng hộ Ken Paxton ở Texas, ông thua. Tổng cộng, các quỹ crypto chi khoảng 176 triệu USD, trong đó khoảng 75 triệu vào hai cuộc đua họ thua.
$BTC gần như không phản ứng, vẫn quanh 77.000 USD.
Tiền nhiều không mua được kết quả. Ngành crypto sẽ phải làm việc với một Quốc hội mà họ đã chi tiền để chống lại.
Bolex 🀄 retweeted
Predict Porto or Napoli, tag a friend and drop your ID before kick-off for a shot at free bets. These UCL promos on Duelbits 1.99 hit different @DuelbitsSports
𝟱 𝘅 $𝟮𝟱 𝗙𝗿𝗲𝗲 𝗕𝗲𝘁𝘀 𝗨𝗽 𝗳𝗼𝗿 𝗚𝗿𝗮𝗯𝘀!
Here's your chance to win a free bet during Porto 🇵🇹 vs Napoli 🇮🇹 in the #UCL tonight 🔥
How to Enter:
✅ Follow @DuelbitsSports
👀 Predict the winner
👉 Tag a friend and drop your User ID
Five winners will be randomly selected from entries submitted before kick-off 🍀
Bolex 🀄 retweeted
Replying to @primus_labs
@primus_labs Epoch 1 ~ rank 287
A lot of projects throw around words like “privacy” and “verifiable data.”
what caught my attention about @primus_labs is that they’re actually trying to build the infrastructure needed to make both work together.
the problem is simple to explain, but harder to solve:
how do you bring useful data from the internet on-chain without exposing all the private information behind it?
Primus is working around two technologies here: zkTLS and FHE.
zkTLS can help prove that data from a website or API is authentic without exposing the underlying information.
FHE takes things further by allowing computations to happen while the data remains encrypted.
so instead of choosing between privacy and usefulness, the idea is:
verify the data → keep sensitive parts private → compute on it → use the result.
that’s what makes the bigger picture interesting to me.
and they’re already applying these ideas beyond just research. Primus has products and infrastructure around things like developer tools, verified profiles, Proof of Reserve, and confidential finance.
the reported numbers are also worth noting:
• 1M+ users
• 10+ paid enterprise customers
• $72M+ in PoR TVL
but for me, the bigger story is where this could go.
on-chain apps are eventually going to need more than wallet balances and transaction history. they’ll need information from the wider internet too.
the challenge is making that information trustworthy without forcing users or businesses to expose everything.
that’s the problem @primus_labs is working on.
getting data on-chain is one thing.
proving it’s real while keeping the sensitive parts private is the much harder part.
Most privacy solutions only address part of the problem.
They either hide everything or reveal everything.
Neither is enough when real capital and real decisions are involved.
What's missing is the ability to prove facts without exposing the underlying data, and to compute on sensitive information without decrypting it.
That's the gap Primus is building to address.
Not more secrecy.
Not more transparency.
But a way to keep sensitive data private while making it usable and verifiable.
Primus is building that cure.
Bolex 🀄 retweeted
The catch was reported as within quota. An independent inspector weighed the haul and found it was well over.
The only observer on board had been paid by the fishing company.
For me, that raises a question beyond skill: how independent is the person assessing the result?
I bring that question to AI model reviews too. If a model grades its own answers, I want a second grader.
That is the lens I brought to @GenLayer's explanation of why GEN exists.
AI judges are cheap to create. If judging carried no cost, one person could try to fill the court with judges they control.
GenLayer requires validators to lock GEN before they can get a vote. Each needs its own stake and infrastructure, making it more expensive to fill the court with controlled judges.
A verdict can also be appealed, bringing in a larger panel.
But I would not call stake proof of independence. It raises the cost of manipulation; it does not guarantee that operators are independent or that a verdict is correct.
When has getting a second opinion changed what you believed about a result?
Bolex 🀄 retweeted
He vouched for the seller. I paid up after barely checking. Then the guy just went quiet. My friend felt bad, and that was the end of it for him, while I was out of pocket.
Certainty is cheap for anyone not on the hook. Since then I weigh every tip by one test: who pays if it fails, and how much?
@GenLayer builds that idea into its validators, which each run a different AI model to rule on cases:
• Each must lock at least 42,000 GEN of its own stake to take on a case
• Cloning a thousand more is nearly free, so a buy-in is the sole way to earn a seat
• Get outvoted by the majority and it forfeits its fee, then hands over an equal sum from that same pool
Words count only once something real is pledged behind them. His carried no backing at all.
If promises required collateral first, would you demand cash, a favor, or their reputation, and why that one?
Bolex 🀄 retweeted
Sitting on a 468x multiplier only to lose $4,675 at the very last moment is wild.
@waxumfriendly, what’s the biggest heart-stopping near-win you’ve seen on @Duelbits? 🤯
Bolex 🀄 retweeted
More creators should know about this.
The @1win Creator Program has a new season running from November 5–14, with $4,000 in rewards for the top 60 creators on the leaderboard.
Here’s how to get involved:
Visit 1w-ugc.pro
Join the Creator Program.
Use my code: R9WYANY
Good luck guys.
🗣️ Jamie Carragher: “It was one of those transfers in the summer where we thought ‘wow, the figure, what was it £117m?” Bargain? Well it is now isn’t it”
“You’d think he scores a certain type of goal but no, his goals are all different” (@SkySportsPL)
Bolex 🀄 retweeted
A dangerous instruction does not need to look dangerous. Sometimes it only needs to be read by the wrong system.
Think of a familiar bank scam. Someone calls, claims to represent your bank and asks for your OTP. The request may sound convincing, but the caller has no authority over your account. You verify through a channel you already trust.
AI agents with wallets face a related security problem: they can process information from public sources and mistake instructions inside that information for commands they are authorised to execute.
The May 4, 2026 Grok and Bankrbot incident showed how serious that boundary failure can be.
The attacker first sent a Bankr Club Membership NFT to the Grok-linked wallet, reportedly granting it elevated permissions. Then they posted a Morse-code message on X and asked Grok to translate it. Grok decoded the hidden transfer instruction and surfaced it to Bankrbot. Bankrbot treated the decoded text as an authorised command and transferred 3 billion DRB tokens. Reports valued the tokens at roughly $150,000 to $200,000.
No private key needed to be stolen for the instruction to trigger a real transaction. The failure happened in how the AI systems interpreted and trusted incoming content.
This is prompt injection: malicious instructions hidden inside content an AI is processing, designed to influence what the system does next.
Greyboxing on @GenLayer addresses this at the input layer. External text is not passed raw to a model. Each validator first screens, paraphrases and restructures the input using its own secret defenses. It then runs its own model. Because validators use different models and independently configured defenses, an attack tailored to one validator is less likely to produce the same result across the group. If one validator accepts a malicious interpretation, the others can outvote it.
That gives an attacker a harder problem than finding one filter that can be fooled repeatedly. They have to contend with multiple validators processing the input through different defenses before the network reaches a decision.
But rewriting introduces its own challenge: remove the hidden instruction without changing the legitimate request.
I would test Greyboxing with a message that asks an agent to summarise a document while hiding a wallet-transfer instruction inside it. Then measure two things: whether any validator still treats the hidden instruction as actionable, and whether the final result preserves the original summarisation request.
What other adversarial test would you add? And what result would convince you that the system is preserving intent without granting authority to untrusted content?
🤝 Paid partnership
Bolex 🀄 retweeted
being transparent onchain shouldn't force you to leak your whole strategy
imagine running an institutional fund where rival desks can literally watch every wallet move you make in real time. your positions, collateral, liquidity and entries all completely doxxed
zero chance serious size wants to trade in an environment like that
this is where @primus_labs gets really interesting
using zkTLS + FHE, they make financial data verifiable onchain without exposing the raw numbers underneath
basically lets big players prove they're good for the money without showing the whole market their hand. huge unlock if we actually want institutional finance onchain
time to cook :s.kaito.ai/M32kFLP
🤝 Paid partnership
public blockchains are great for transparency, but most of the time they end up exposing way too much user info
@primus_labs tackles this by combining zkTLS for verifiable web data with FHE to run computations directly on encrypted data, already powering private DeFi use cases over on BNB Chain
put together this quick walkthrough to break down how all the pieces fit together. curious what you guys think should always stay private onchain?
time to cook :s.kaito.ai/M32kFLP
My test for any system that decides things about people: who gets a second look when the first one is wrong?
Anthony Ray Hinton's first look was a lawyer who believed he had $1,000 to hire an expert. The bullets never matched his gun. He spent nearly 30 years on death row.
What I can't shake is how ordinary the mistake was: a wrong guess about a budget.
Agents are about to make millions of deals. When one goes wrong, the first look will be a model's too. I'd want the second look built in before the deal, not argued for afterwards.
@GenLayer is built so no single reviewer decides alone.
When did you last catch an AI being wrong before it cost you?
#JusticeNeedsAnUpgrade
Slow. Expensive. Wrong.
It's been this way for centuries. It doesn't have to be.
#JusticeNeedsAnUpgrade
Bolex 🀄 retweeted
Đi học từ chuyện “quyền” thành “nghĩa vụ”. Mình nghĩ ý tưởng tốt, nhưng quan trọng nhất là có những em nào vẫn chưa đến trường vì hoàn cảnh, và ai sẽ giúp các em? Mọi người nghĩ sao?
Bolex 🀄 retweeted
Sometimes I look at how far humanity has come and still wonder how much we have yet to discover.
We’ve made incredible progress, but there’s still so much about the universe we barely understand.
That’s one reason @projectspherexy keeps my attention.
The idea behind Sphere makes you think beyond what we have today and imagine what humanity could achieve if we keep pushing the limits of energy and technology.
There’s still so much ahead of us, and it makes me wonder: Do you think humanity will ever reach Kardashev Type II, or is that still far beyond what we can achieve?
I’d love to hear what you think.
Gm Ct ☀️❤️
It's a new week filled with opportunities and wins 🤲🏼🎉
Stay active to have a share in it 💰🔥





































