Turning data into profit 📈 Web3 | DeFi Strategist Airdrop Hunter Open for Collab 🤝 | DM for Ads 📩
Joined April 2026
- Tweets536
- Following6.4K
- Followers3K
- Likes779
Æli Jæbeer retweeted
A polite favor, a chatbot, and then an agent handed over around $200,000. The weapon was a plain translation request.
An attack like this needs all of these to go right:
• it reads what comes in
• it trusts the content
• it acts and funds move
The May 2026 case tucked a payment note inside Morse and made Grok put it in English. It complied, passing the decoded words to Bankrbot, a program that turns X posts into onchain transactions, and it obeyed: the tokens left. Press figures start near $150,000, and the video rounds up.
That ploy is called prompt injection: a command hidden in material the system should see as data. Crypto is not alone in this. It sounds rare, yet any bot that checks inbox mail, social replies or web visits shares the risk. A toy issue, until it can pay.
At the front anyone could speak; the middle was a blind handoff; at the end a purse opened, behind a lone door to study.
@GenLayer’s Greyboxing cuts the chain at its weak spot. Nothing a sender types arrives at the AI unaltered. Each validator begins by:
• vetting the message
• rewording it
• reshaping it
Then it uses private defenses of its own, plus a separate model. An exploit built for a solo target becomes static on all the others, and any that still falls gets overruled.
No setup is ever airtight this way. The aim is small: a duped judge must not settle the result.
Think of the tool you lean on daily. Which job bugs you most?
• taking in input
• believing it
• paying out
Reply with a choice and why.
🤝 Paid partnership
Æli Jæbeer retweeted
The apartment was perfect. That's exactly why I lost my deposit.
It was in Ughelli, Delta State. The property agent walked me through fresh paint, working taps, clean tiles. I was sold in ten minutes.
I paid. Then it rained.
Behind the wardrobe, water was pouring through the ceiling.
The deposit was non-refundable. I'd been shown exactly what someone wanted me to see, on the one dry day I looked.
What if I'd brought five inspectors, each checking in their own way, without telling the seller what to look for?
You can't stage a place for that.
That's the part of @GenLayer's Greyboxing I find most interesting.
Before any model sees your input, every validator rewrites it in its own private way. Then each runs its own model.
A trick built for one model can fall flat on the rest. If one still bites, the others outvote it.
Why does that matter? AI agents that hold money can be shown a staged scene too.
In May, someone smuggled a transfer instruction into an X post using Morse code. Grok decoded it, Bankrbot took the result as an order, and around $200,000 moved.
That's prompt injection: an instruction hidden inside content the AI was only meant to read.
Same trap, just with more zeros.
GenLayer's video runs one minute. My apartment viewing ran ten, and it still missed the leak.
What's the one thing you'd never let an AI agent pay for without a second opinion?
Æli Jæbeer retweeted
My first three rooftop tomato plants drowned because I treated watering like an anxious chore rather than a rhythm.
I dumped a full can every morning convinced more moisture meant faster fruit. I got waterlogged roots and yellow drooping stems instead.
For two weeks I kept a simple kitchen window log. Twenty four pots tracked by hand. Every entry recorded soil moisture by touch and the exact weather before I lifted the spout. Pot 4 died because I ignored drainage holes. Pot 12 struggled because I watered on a panic schedule.
Somewhere around pot 15 the habit shifted. I stopped guessing and started checking knuckle deep before adding a single drop. The logbook stopped being a list of dead stems and became a reliable filter for patience.
Last month a neighbor asked why my patio survived the dry spell. I heard myself repeat the golden rule the pots taught me: dry soil teaches roots to reach deeper.
That lesson is why @RallyOnChain resonates with what I build. Real persistence values quiet daily attention over loud speed. Earned, not announced.
What habit did you finally stop forcing once you learned to test the depth first?
🤝 Paid partnership
Æli Jæbeer retweeted
Every leverage product sends a bill. Almost none tell you who really pays it.
Hidden funding rates, swap fees, a slow decay: the invoice exists, just buried.
@2FactorFinance shows its books.
• Risk is cut into two perpetual tranches
• Senior: the shielded slice, where the junior's fee winds up
• Junior: the boosted portion, with no liquidation trigger and nobody hedging against it
Common setups are priced off shorting costs. Here the basis is stable yield, cheaper and calmer.
Each asset receives a custom multiple of its own from its trend speed and how wildly it swings. Bitcoin, near 60 percent volatility, lands right around 1.33x; stocks, at 16 to 19%, run past 3x. Base is the home; the crypto leg is on testnet.
Season One, in plain terms:
• Earned through checked actions: writing, studying, inviting
• Balances score zero too
• The program closes at launch, freezing the board
• The leading ten ranks share 1 BTC, paid as cbBTC: 18.2% for first, 1.8% for tenth
Marks hold no cash value, can't move across accounts and do not represent ownership of any token.
What I'm watching: whether the geared side survives a long slump. That is the only test that matters.
Drop a single number: the most gearing that coin can carry for a decade without fading. The design already has one; poke it. Best case earns a reply from me.
My invite (I benefit if you qualify):
points.2factor.finance/r/su8…
Æli Jæbeer retweeted
My boyfriend told me a ₦25,000 power bank cost ₦45,000. He bet I wouldn't check.
I checked.
An AI agent with a wallet doesn't check. It reads, then obeys. In May, that cost around $200,000.
Someone hid a transfer instruction in Morse code in an X post. Grok decoded it and passed it to Bankrbot, which moved the tokens.
Grok saw a translation job. Bankrbot saw an order.
That's prompt injection: a hidden instruction the AI obeys.
A single model behind a single filter, with all the time in the world for an attacker to practise on it.
On @GenLayer, Greyboxing flips that. Every validator rewrites the input with its own secret defences before its own model sees it.
A trick built for one model's weak spot may do nothing to the rest. If one validator still falls for it, the others can outvote it.
Far less to study, far harder to practise on.
What would you want standing between your agent's wallet and a stranger's tweet?
Æli Jæbeer retweeted
Owing $100 meant nothing on day one. A few very quiet days later, it was everything.
A friend lent it for stocking my shop. I expected quick sales, so paying him back seemed a given.
Then customers asked prices and left empty-handed. I checked the till daily. Even a small gain came with the debt as my first thought. He never pushed me, but I carried it.
Being correct about the idea was not the problem. Having room to wait for it was.
That is the lens I used to read @2FactorFinance. It runs on Base, works with Coinbase, and brings Kelly-optimized leverage onchain, a fit for Coinbase's tokenized stocks. Two forces work against leveraged holders over time:
• volatility drag, which is pure math
• the cost of the capital behind the exposure
Most such structures get liquidated midway or decay, so they suit only short spans. 2Factor is built to hold through the delay.
Season 1 of the Points Program rewards verified actions:
• social activity
• education
• referrals
Putting money in earns no Marks. They have no cash worth and are not a claim on any token. The opening round ends at launch, when the board freezes. Ten ranks split 1 BTC, paid in cbBTC.
Open my link, finish one task, and see where you stand tomorrow:
points.2factor.finance/r/fbg…
What are you holding at this moment that could not survive a slow month? Name it, and say how many weeks it might last.
Æli Jæbeer retweeted
A few years ago, I joined a small community savings group with some friends.
Everyone was excited at the beginning. We all contributed, shared responsibilities and agreed that everyone would benefit according to the rules.
But there was one problem: nobody had properly explained how the contributions would be tracked or how the final allocation would work.
By the time questions started coming up, people had different interpretations of the original agreement.
That experience changed how I approach reward programs. Before participating, I want to understand three things: what qualifies as participation, how progress is measured, and how rewards are distributed.
That's partly why I took an interest in @2FactorFinance.
I've joined its Points Program, and the structure is straightforward.
Season 1 rewards verified actions across three categories:
• Social activities: participating in discussions and sharing relevant content.
• Education: learning about the protocol and completing educational activities.
• Referrals: introducing other people to the ecosystem.
One important distinction is that buying, depositing or holding assets does not earn Marks. Participation is based on verified actions.
The leaderboard also has a defined reward structure.
When Season 1 ends at the launch of 2Factor Finance, the leaderboard freezes. The top 10 accounts share 1 BTC, paid in cbBTC, according to a fixed allocation:
Rank 1 receives 18.2%, with the allocation decreasing to 1.8% for rank 10.
Marks themselves have no cash value, cannot be transferred and do not represent a claim on any token or asset.
But there's another reason I'm following the program beyond the leaderboard.
2Factor Finance is building on Base and working alongside Coinbase to bring Kelly-optimized leverage onchain, with an interesting application for tokenized stocks.
Its approach considers the volatility and long-term drift of individual assets when determining appropriate leverage, rather than applying the same multiple everywhere.
The underlying structure separates exposure into a senior tranche with deep downside protection and a leveraged junior tranche without liquidation risk or an external hedging counterparty.
I find the connection between the product and the Points Program interesting. The program encourages people to learn, discuss and introduce others to a financial mechanism that deserves proper examination.
I'm participating in Season 1, and I'll be following how the ecosystem develops.
If you want to participate too, you can join through my referral link:
points.2factor.finance/s/h44…
One thing I'd like to ask: when joining a points program, what matters more to you: transparent reward distribution, meaningful activities, or the actual product behind the campaign?
Æli Jæbeer retweeted
went hunting for the catch in this campaign. Traps hide in three places; I checked each.
•Cost of entry: effort only. Posting, studying and referring are credited as Marks once verified; buying or parking idle funds adds zero.
•Reward: Season One ends at launch, freezing the ranking. Ten accounts share one BTC (cbBTC), from 18.2% for first place to 1.8% for tenth.
Those points can't be cashed out, moved, or redeemed for anything.
•Engine: @2FactorFinance, native to Base, working with Coinbase, sizes leverage by the Kelly criterion for tokenized stocks. Volatility tied to every asset sets the gearing, so Bitcoin lands near 1.33x rather than a flat 3x.
Putting stocks onchain fixed distribution and skipped structure. This is an attempt to add it.
Link (qualified entries credit me):
points.2factor.finance/r/1oi…
Your move: name the rule you'd demand from a scheme. I'll test this one against it live in the replies.
Æli Jæbeer retweeted
A friend of mine runs a small delivery business. He started with one motorcycle, then bought three more when demand increased.
At first, expansion looked like the obvious decision. More motorcycles meant more deliveries and more revenue opportunities.
But he overlooked something: every additional motorcycle came with fuel, maintenance, repairs and operating costs. More capacity did not automatically mean better efficiency.
That experience taught me something about leverage. Increasing exposure is one thing. Structuring the cost of that exposure is another.
This is part of what makes @2FactorFinance interesting to me.
Its public beta is live on Base, working alongside Coinbase to bring Kelly-optimized leverage onchain, with a particularly interesting application for the tokenized stocks ecosystem.
Traditional leveraged products often struggle with two compounding forces: volatility drag and financing costs. A position can move in the expected direction over time and still suffer from the structure used to maintain that exposure.
2Factor approaches this through volatility segmentation, separating exposure into two perpetual tranches: a protected senior side and a leveraged junior side without liquidation risk or an external hedging counterparty.
The leverage multiple is also asset-dependent. The research places Bitcoin around 1.33x, gold around 2.35x and the S&P 500 around 2.1x. Different assets have different volatility profiles, so applying one fixed multiple to everything misses an important part of the equation.
I'm also participating in the 2Factor Points Program.
Season 1 rewards verified actions across three areas:
• Social activities
• Education
• Referrals
No buying, depositing or holding is required to earn Marks.
There's a 1 BTC prize pool for the top 10 leaderboard participants when Season 1 ends at launch. Distribution is in cbBTC, following a fixed rank-based allocation.
One detail worth understanding: Marks have no cash value, cannot be transferred and do not represent a claim on any token or asset.
I'm more interested in understanding how this mechanism handles long-term exposure than simply looking at the leverage multiple.
Join the program through my referral link:
points.2factor.finance/s/h44…
When evaluating leverage for a long-term position, do you pay more attention to volatility or the cost of financing?
Æli Jæbeer retweeted
A troll comment buried in a PDF once got my AI assistant to delete notes I never asked it to touch.
The file was meant to be read, not obeyed. Give that same weakness access to a wallet, and the consequences get expensive fast.
In May 2026, someone hid a transfer instruction in Morse code in an X post. Grok decoded it and passed the text to Bankrbot, an agent that executes onchain transactions from X posts. Around $200,000 in tokens moved.
That's prompt injection: a command smuggled into content an AI is only supposed to read.
@GenLayer's Greyboxing rewrites incoming text before any model sees it. Each validator screens and restructures the input using its own secret defenses, then runs its own model. If one is fooled, the others can outvote it.
How much should validators rewrite incoming text to block hidden instructions without changing what the user actually meant?
Æli Jæbeer retweeted
I study chemical engineering, and the first thing you learn about any process is that it has an operating window. Too little pressure and nothing reacts. Too much and the vessel fails. The skill is knowing where the window sits, and the material sets it, not ambition.
Leverage has a window too. Two forces compound against anyone who holds it: volatility drag, which is mathematical, and financing drag, the cost of the capital behind the exposure. Where an asset's long-run drift outpaces both, there is a band of leverage ratios where compounding works for the holder instead.
That is the idea behind @2FactorFinance, whose public beta is live on Base. It starts with Bitcoin, where volatility near 60 percent closes the band early and the target is about 1.33x, and it works with Coinbase on Kelly-optimized leverage for tokenized stocks next.
Season 1 of the Points Program works like this:
Marks come from verified social actions, education and referrals.
Buying, depositing or holding earns none.
Marks have no cash value and cannot be transferred.
When the leaderboard freezes at launch, the top 10 split 1 BTC, paid in cbBTC, from 18.2 percent at rank 1 down to 1.8 percent at rank 10.
My link, if you want to run the process yourself:
points.2factor.finance/r/5jx…
What field do you know well enough to name its operating window?
Æli Jæbeer retweeted
Every product got the same $200 slice. Simple, fair, and wrong. That was my restock plan, and it cost me.
Weeks later the truth showed up on my own shelves: three items vanished daily, the rest gathered dust. Money stayed frozen in what nobody wanted, while best sellers ran dry.
Oddly, that mess is the cleanest lens I use for @2FactorFinance. It lives on Base, teams up with Coinbase, and brings Kelly-optimized leverage onchain, a fitting match for Coinbase's tokenized stock world. Instead of one number for all, each asset gets the size its own behavior can bear:
• Bitcoin, at roughly 60% volatility: about 1.33x
• S&P 500, at 16 to 19%: about 2.1x
• Gold: about 2.35x
Inside, an asset's swings are cut into two tranches: a senior one with deep downside protection, and a leveraged junior one with no liquidation risk.
Season 1 of the Points Program pays only for verified actions:
• social posts
• education tasks
• referrals
That is the whole list. Nothing you put in earns Marks, and they are no claim on any token.
Ten leaderboard spots divide 1 BTC, paid in cbBTC, on a fixed curve from 18.2% for first place down to 1.8% for tenth.
Pick one task, post it today, and see where you land. My link:
points.2factor.finance/r/fbg…
Which shelf in your life is stuffed with slices that never earned them? Reply with yours, and I will tell you what I would cut back.
Æli Jæbeer retweeted
The most dangerous AI in the $200,000 theft was not the one holding the money.
It was the translator.
Someone hid a transfer order inside Morse code and asked Grok to translate it. Grok did its job perfectly. It decoded the message, cleaned it up, and handed it to Bankrbot as plain text. Bankrbot read an instruction in what looked like an ordinary message and moved around $200,000 in tokens.
Nothing was hacked. A helpful tool carried a stranger's command across the border, word for word, because carrying things faithfully is what it is for.
That is prompt injection when the reader controls a wallet. The payload survives because every step on the way was built to preserve it.
This is the part of Greyboxing I keep coming back to. On @GenLayer, user text never reaches a model in its original form. Every validator rewrites the input first, screening it, paraphrasing it, restructuring it, each with defenses it keeps secret and a model of its own.
A smuggled instruction is written for one reader and tuned against one system. After rewrites by validators it cannot study, whatever arrives is a rumor of the original. Rumors do not move wallets.
And if one validator still falls for what is left, the rest outvote it.
One lock can be studied until it opens. A lock that changes shape in secret gives an attacker nothing to rehearse against.
You paste text written by strangers into AI tools every day. We all do. What would change if every tool treated that text as something to rewrite first, instead of instructions to obey?
Æli Jæbeer retweeted
Leverage has two kinds of customers. People who trade it for a few days, and people who'd like to hold it for years.
Nearly every product is built for the first group. Daily resets. Flat 2x and 3x multiples. A liquidation price that closes you out on one bad hour. That's reasonable if you're in and out by Friday, and wrong if you plan to hold.
2Factor Finance is built for the second group. It runs on Base and is aimed at Coinbase's tokenized stocks, and it makes three different choices.
The multiple is set per asset with Kelly sizing, so a more volatile asset gets less leverage. For cbBTC that comes out around 1.33x, not a flat 2x.
There is no liquidation price, so a single hour of volatility can't end the position.
The extra exposure is financed by investors who'd rather hold the safer side of the same asset, not by traders betting against you. The launch announcement puts the cost of carrying it at a fraction of a leveraged ETF's.
I joined the 2Factor Points Program. Season 1 rewards completed actions in three categories: social activity, education, and referrals. Balance and deposit size don't count for anything.
Education fits this product well. The sizing logic is easier to judge once you understand why 1.33x and not 2x, and what changes when there's no liquidation price.
You complete actions, earn Marks, and move up the leaderboard. At the end of Season 1, the top 10 accounts split 1 BTC. Marks carry no cash value, can't be transferred, and aren't a token. They're a record of what you did during the season.
My referral link, if you want to see the task list: points.2factor.finance/r/8a8…
@2factorfinance
When you've held leveraged exposure, which part cost you most: the multiple, the carry, or the forced exit?
🤝 Paid partnership
Æli Jæbeer retweeted
Losing my project laptop was the price I paid before I finally took my gambling problem seriously.
At school, gambling had become what people associated with my name. My dad's warnings sounded like background noise until that loss forced me to reconsider my choices. I knew something had to change, and I started taking responsibility for it: • I began visiting counsellors. • I distanced myself from certain friends. • I kept working on myself until gambling was no longer part of my life.
Today, that chapter is behind me. That's why @RallyOnChain resonates with me. It recognises that genuine effort and quality matter more than status or where someone starts. A bad chapter shouldn't have the final say in your story.
What's something you once struggled to walk away from that changed the direction of your life?
Skip the prize and read the fine print first. Here is what a points program owes you before it earns your time, checked against 2Factor Finance.
Cost to enter: nothing. Points come from verified activity only: posting, learning, inviting. Spending or parking money counts for zero.
Reward on the table: when Season 1 ends, the ten highest ranks split 1 BTC. The points themselves carry no cash value and cannot be moved or sold.
Product behind it: @2FactorFinance on Base, offering leverage sized by Kelly math for each asset separately, tied into Coinbase's tokenized stocks, rather than a single flat multiple for everything.
Nothing hidden beyond that.
I signed up to watch how the product develops, and if someone qualifies after clicking my link, I may collect points too:
points.2factor.finance/r/biq…
What is the one thing you would demand a points program answer before you spent a single minute on it?
Æli Jæbeer retweeted
Ten seconds left and I would grab the external hard drive I have not opened in four years.
Not the phone. Not the passport. The drive that still has every photo from a phone I threw away in 2019.
I do not look at them. I just cannot leave them behind.
The filter is simple: if losing it would force me to explain a gap I cannot fill, it comes with me.
@RallyOnChain what is the one useless thing you would still grab in ten seconds?
🤝 Paid partnership
Æli Jæbeer retweeted
Most seasons end on a date. Season 1 of the @2FactorFinance Points Program ends on an event: the launch. When 2Factor Finance launches, the leaderboard freezes, and the ranking at that moment is the ranking that counts.
That design choice tells you what the program is for. It ties the standings to the product instead of a calendar, and in Season 1 the standings come from verified social actions, education and referrals.
Here is the product those actions are about. 2Factor Finance builds leverage for people who intend to hold, and its public beta is live on Base. Two forces compound against leveraged holders: volatility drag, which is mathematical, and financing drag, the cost of the capital behind the exposure. The response is to split an asset's volatility into a senior side with deep downside protection and a junior side with no liquidation risk, and to set the multiple from the asset itself. For Bitcoin that is about 1.33x. It also works with Coinbase on bringing Kelly-optimized leverage to tokenized stocks.
What the freeze decides:
1 BTC, paid in cbBTC, split across the top 10
Rank 1 takes 18.2 percent, rank 10 takes 1.8 percent
Marks have no cash value, cannot be transferred and are not a claim on any token
If you want a place on that board, start here:
points.2factor.finance/r/jgg…
Would you rather a season be tied to a date or to an event, and why?
Æli Jæbeer retweeted
A climbing BTC chart almost had me borrowing to buy in. Then it reversed as I watched.
The setup looked promising and the dip was small. One slight slip the wrong way proved borrowed exposure squeezes fast. I quit predicting market direction and began asking how much damage a position can absorb.
That shift is why @2FactorFinance caught my eye. Live on Base and teaming with Coinbase, it brings Kelly-optimized leverage onchain, the missing structure for tokenized stocks. The multiple fits the asset's volatility: about 1.33x for Bitcoin, well above 3x for calmer equities. The geared side faces no liquidation.
I joined the Points Program, where Marks come only from verified actions:
• posting on X
• studying the math behind it
• referring others
Buying, depositing or holding earns zero. The tally isn't currency, can't go to another wallet, and entitles nobody to a token. When Season 1 ends at launch, the top 10 share 1 BTC in cbBTC.
My referral link (I benefit if you sign up):
points.2factor.finance/r/lab…
Which trade nearly pulled you in, and what kept you out? Give the asset and the number. Best stories win my longest response.
