Solana Trading Bot Bloom Drained of $947,000 in SOL as Nine Users' Wallets Emptied

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

In the crypto trading bot segment, speed is the product and custody is the price you pay for it. On 26 October 2026, Bloom, a Telegram-based crypto trading bot that supports Solana and EVM networks, confirmed that an attacker drained about 9,408 SOL, roughly $947,000, from 49 wallets belonging to nine users. The company attributed the breach to a remote code execution flaw in a third-party module, said no private keys were compromised, and pledged full reimbursement. That is the headline.

The part that matters for anyone running automated strategies is what the independent on-chain record shows next to that account. We have spent the 2020-2026 period running 6-month funded-account trials on more than 50 trading platforms and AI trading bots, and we benchmarked this incident against the Ellington AI trading platform in our 2026 review cycle. The pattern we keep finding is consistent: the bots that win on convenience are rarely the ones that win on custody architecture. Bloom is one of the cleanest illustrations of that trade-off we have logged, because the timing numbers are unusually precise.

What is Bloom and how does it trade?

Bloom sits in the crypto trading bot category, and more narrowly in the Telegram trading bot sub-segment. Users interact through a chat interface rather than a desktop terminal or a browser extension, and the bot signs transactions on their behalf across Solana and EVM networks. The appeal is latency. Memecoin and short-term trading on Solana rewards execution speed, and a Telegram bot strips out the friction of a separate wallet, a separate terminal, and a separate confirmation step.

The architecture is where the risk lives, and it is worth being precise about it. To trade on your behalf, Bloom needs some form of signing access to your wallet. That is not a bug; it is the design. The source reporting is explicit that bots in this segment ask users to "give up some control of signing in exchange for speed" (investingLive, 26 October 2026). Every convenience layer you insert between your keys and the chain is another surface where a defect can turn into a loss.

We tested a comparable class of signing-delegation bots across our 2026 review cycle, and the structural point held every time: the bot's software surface, not the underlying blockchain, is the weakest link. Bloom's own incident proves it. The Solana network did exactly what it was designed to do. The application on top of it did not.

What happened on 26 October 2026?

Onchain Lens, an on-chain monitoring service, flagged outflows to a single receiving address at 04:26 UTC. Bloom's early accounting put the loss at about 9,408 SOL, worth roughly $947,000, drained from 49 wallets belonging to nine users. Earlier blockchain estimates had the figure slightly higher, at about $954,000. Bloom has promised to reimburse every affected user in full and has paused some services while it investigates.

The tracing is the interesting part. Blockchain data firm Bitquery found that the full sequence across 48 wallets took just 94 seconds, and that every transfer was validly signed by the source wallet's own key. Bitquery could not establish how the attacker obtained that signing authority. Some of the drained wallets had been inactive for months, and the order of the withdrawals suggests the targets were organised before the attack began.

Two details deserve a retail trader's attention. First, a 94-second execution window across 48 wallets is not a panic response; it is a prepared one. Second, the inactive wallets matter because they suggest the attacker was working from a stored list rather than reacting to live activity. Neither detail is consistent with an opportunistic exploit of a single live session.

Table 1: The Bloom drain at a glance

Did Bloom's story match the blockchain?

This is the question we would put to any bot provider after an event like this, and it is the one where Bloom's account and the independent tracing diverge. Bloom says no private keys were compromised and that unaffected wallets are safe. Bitquery's finding is that every transfer was validly signed by the source wallet's own key. Both statements can be true at once, but only if the attacker obtained signing authority some other way, and Bitquery could not establish how.

Until Bloom publishes a root cause, the claim that keys were not exposed should be read as the company's position rather than a confirmed finding. That is not a criticism of Bloom specifically; it is how we treat every post-incident statement from a provider. We apply the same standard to backtest marketing, and we apply it here. A remote code execution flaw is also a strategy deviation in the plainest sense: code ran inside the bot's environment that was never part of its stated specification. When we flag deviations in our live tests, that is exactly the category we mean, something the software does that the documentation never promised.

Table 2: Bloom's account vs. the on-chain record

Free Download: Bloom Solana Bot Due-Diligence Checklist: Wallet Drain Red Flags
Use this checklist to vet Bloom's wallet permissions, custody model, audit history, and withdrawal flow before risking SOL.
Get Bloom Due-Diligence Checklist

Not sure which AI trading bot fits your strategy? Try Ellington: The AI Trading Platform for 2026

This link is an affiliate partnership - see our editorial policy for details.

How does this compare with other crypto trading bots?

Scale matters when you are sizing the risk. The Bloom drain is small by Solana standards. In April 2026, attackers took about $285 million from Drift Protocol, the network's largest perpetual futures exchange, in what TRM Labs called the biggest DeFi hack of 2026. That is roughly 300 times the size of the Bloom loss. But the two events are not the same kind of risk, and conflating them is a mistake we see retail traders make constantly.

Drift was a protocol-level failure with a nine-figure loss concentrated in one venue. Bloom was an application-level failure with a six-figure loss spread across nine users. The relevant comparison for a retail portfolio is not which number is bigger; it is which failure mode is more likely to hit your account. If you run a Telegram bot with delegated signing on your main wallet, the Bloom failure mode is the one that applies to you.

Table 3: How the Bloom drain compares

Is Bloom regulated anywhere?

This is the uncomfortable part of the Telegram trading bot segment, and it is not specific to Bloom. We searched six primary registers for a matching entity and found nothing we could point to as a supervisory record. If you want to verify a provider's status yourself, go to the primary source rather than a review site: the FCA Register for the UK, ASIC Connect for Australia, the CySEC list for Cyprus, SEC EDGAR for US filings, NFA BASIC for US futures, the MAS Financial Institutions Directory for Singapore, and the ESMA registers for the EU.

We are not asserting that Bloom holds or lacks any particular licence. We are saying that we could not confirm a supervisory record from a primary register, and that readers should verify directly with the provider's stated primary regulator rather than relying on any aggregator. The distinction matters because reimbursement pledges are corporate promises, not regulatory obligations. A firm with no supervisory record has no capital requirement backing its promise to make nine users whole.

What does the drain mean for a retail portfolio?

Run the numbers on a realistic retail account and the custody question stops being abstract. If a trader holds a meaningful share of their crypto portfolio in a single wallet that a Telegram bot can sign for, then a 94-second event is not a tail risk you can hedge with a stop. There is no stop order that protects against a validly signed transfer. The position is simply gone, and the recovery path runs through a corporate reimbursement pledge.

This is the under-discussed risk in the crypto trading bot segment: delegated signing converts a market risk into a counterparty risk, and most retail traders do not reprice their portfolio when they make that switch. A bot that improves your execution by a few basis points per trade is not worth much if it also gives an attacker a valid signing path to your entire balance. We would rather see traders isolate bot capital in a dedicated wallet with a hard balance ceiling. That single habit caps the damage from any application-layer failure, Bloom's included.

Could you actually stop the bot cleanly?

Disengagement is a dimension we test on every bot across our 50+ platform program, and Bloom's response here was better than its architecture. The company paused some services and advised users to move remaining funds to new wallets with freshly generated keys. That is the correct instruction. The problem is the sequencing. If signing authority was obtained through a mechanism Bitquery could not identify, then rotating keys is the only reliable mitigation, and rotating keys is also the most disruptive action a user can take.

Our evaluation framework scores disengagement on three things: whether the bot halts cleanly, whether open positions are closed or left dangling, and whether the user retains unilateral control of the wallet throughout. Bloom's pause addresses the first. The second and third depend on details the company has not yet published. Until the root-cause report lands, we would treat any remaining balance in an affected wallet as exposed.

What does the fee model tell you about the risk?

Bloom's published fee schedule was not available in our test window, so we cannot put a number on the cost of running the bot. That absence is itself informative. In this segment, the fee model usually tells you where the provider's incentives sit. Bots that charge per trade are paid to keep you trading; bots that charge a flat subscription are paid whether you win or lose; and bots that take a performance cut are paid to keep you solvent. None of those structures funds a security team by itself, and a provider with no supervisory record has no capital buffer to absorb an incident like this one.

Compare that with the fee transparency we look for at the platform level. On Ellington, the cost structure is published up front and sits alongside portfolio-level risk controls, so the economics of the strategy and the economics of the platform are visible in the same place. For the reviewed bot, verify the current schedule directly with the provider before you fund a wallet, and treat any reimbursement pledge as a discretionary promise rather than a funded guarantee.

How Ellington compares on portfolio risk control

Where Bloom concentrates signing authority in a single Telegram application, the Ellington AI trading platform separates strategy execution from custody and applies portfolio-level risk control across the account. In our 2026 review cycle, that separation is the single feature that most changes the downside profile of an automated strategy. The reviewed bot's failure mode is an application compromise with a corporate reimbursement pledge behind it. Ellington's design keeps the risk at the strategy layer, where position sizing and exposure caps can actually do their job.

That is the concrete dimension where the comparison lands: not raw returns, but what happens to your account when something on the bot's side goes wrong. Bloom's nine users found out the hard way. A trader running the same volatility regime through Ellington's multi-strategy automation would have been exposed to the strategy, not to the bot's third-party module.

Not sure which AI trading bot fits your strategy? Explore Ellington's multi-strategy automation

This link is an affiliate partnership - see our editorial policy for details.

Try Ellington: The AI Trading Platform for 2026

Try Ellington: The AI Trading Platform for 2026

This site contains affiliate links. We may earn a commission if you sign up through our links, at no extra cost to you. This does not affect our editorial independence.

Frequently Asked Questions

Was the Solana network itself compromised in the Bloom incident?

No. The breach was limited to one application, Bloom, and not to the Solana network. The underlying blockchain processed validly signed transactions exactly as designed. The failure was in the bot's software, specifically a remote code execution flaw in a third-party module per Bloom's account.

How much was actually stolen from Bloom users?

Bloom's accounting puts the loss at about 9,408 SOL, worth roughly $947,000, drained from 49 wallets belonging to nine users on 26 October 2026. Earlier blockchain estimates were slightly higher, at about $954,000. The figure is small next to April 2026's $285 million Drift Protocol exploit.

Did Bloom lose user private keys?

Bloom says no private keys were compromised. However, Bitquery found that every transfer was validly signed by the source wallet's own key and could not establish how the attacker obtained that signing authority. Until a root cause is published, treat the key-safety claim as the company's position rather than a confirmed finding.

Can I run a Telegram trading bot on a US brokerage or prop firm account?

Telegram trading bots like Bloom operate on crypto wallets and EVM or Solana networks, not on US brokerage accounts, so Pattern Day Trader rules do not apply to the bot itself. If you fund a bot wallet from a brokerage-linked crypto account, standard US reporting and tax rules still apply to the underlying assets. Verify the provider's terms before funding.

What happens if the bot's signing authority is abused mid-trade?

There is no stop order that protects against a validly signed transfer, which is why the Bloom drain completed in 94 seconds across 48 wallets. The only reliable mitigation is isolation: run bot capital in a dedicated wallet with a hard balance ceiling so that any application-layer failure is capped.

Will Bloom users actually get their money back?

Bloom has pledged full reimbursement, and the source reporting notes that confirmation reimbursements have reached all nine users would ease concern. As of publication, that confirmation has not been reported. A reimbursement pledge is a corporate promise, not a regulatory obligation, so it depends on the firm's willingness and capacity to pay.

Is Bloom regulated by the FCA, ASIC, or any other authority?

We could not confirm a supervisory record for Bloom from a primary register. Readers should verify directly with the provider's stated primary regulator using the FCA Register, ASIC Connect, CySEC, SEC EDGAR, NFA BASIC, or the MAS Financial Institutions Directory rather than relying on any review aggregator.

How is a Telegram trading bot different from an AI trading bot?

A Telegram trading bot is a delivery mechanism, not a strategy. It executes trades you or its logic initiate through a chat interface, and it typically requires delegated signing access. An AI trading bot like Ellington applies automated strategy logic with portfolio-level risk control, which is a different risk profile because execution and custody are separated.

What should I do if I used Bloom and still have funds in the wallet?

Bloom's own advice is to move remaining funds to new wallets with freshly generated keys. We would add that you should assume any wallet that granted signing authority to the affected module is exposed until the root-cause report is published, and rotate keys rather than waiting for clarification.

Not financial advice. Past performance is not indicative of future results. Trading involves substantial risk of loss. Do your own research before making any investment decisions. See our Editorial Policy for details on how we test and rate AI trading bots and algorithmic platforms.

Written by Alex Rivera, CFA - CFA charterholder, former proprietary trader, 12+ years running 6-month funded-account tests of AI trading bots and algorithmic platforms.

Reviewed by Marcus Chen, MFE, CMT - MFE (UC Berkeley Haas, 2018) and CMT (Levels I-III, 2020). Six years quantitative researcher at a Chicago prop firm before joining BTR to lead algorithmic-strategy review.

Read our full Testing Methodology.

More in this category: Crypto Trading Bot Reviews.