Core DAO: CORE Transfers Frozen, Emergency Hard Fork Announced. What Our Chain Measurement Shows

If you hold CORE, one question currently comes before the price: the question of where it sits. If your holding is on an exchange, it may have been stuck since 31 August, depending on the provider. Coinbase has paused deposits and withdrawals on the Core DAO network, and several other trading venues have followed. The project itself speaks of validators that received markedly more reward than intended, and is preparing an emergency hard fork.

We therefore went beyond the reports and queried the chain itself. On 2 September 2026 we retrieved and analysed around 13,000 blocks of the Core mainnet individually. The result yields three numbers that appear in no report: the number of block-producing addresses rose from 21 to 23 on 28 August and fell back to 21 on 31 August. In the early hours of 31 August the chain lost 181 blocks against its otherwise exact cadence. And at no point during that period did it stand still for longer than five seconds.

What happened at Core DAO and who the transfer freeze affects

Core DAO is an independent blockchain with its own token, CORE. Unlike a token that lives on Ethereum, it runs its own validators, which build blocks and are rewarded for doing so. That is precisely where the incident sits.

On 31 August 2026 the project disclosed that a small number of validators had accumulated block rewards well above the issuance provided for in the protocol. A day later, on 1 September, Core stated that the matter had been contained and that the validators concerned could no longer draw excess rewards. According to the project, the incident affected only the issuance of rewards; user balances were said to be untouched. To close the fault permanently, Core is coordinating a hard fork with the validators.

You are affected in the first instance if your CORE sits with a provider that has paused deposits and withdrawals on the Core network. You are affected in the second instance if you hold CORE in your own custody and want to move it over the coming days: in that case you should know whether and when the hard fork takes effect. What is not affected is your holding as such, because a forward upgrade does not reverse confirmed transactions.

Validator, block reward and hard fork: the three terms in one sentence each

A validator is a machine with staked capital behind it that is entitled to propose and confirm new blocks, and is paid for doing so. The block reward is the amount the protocol pays out to those validators per block; its size is fixed in the software and is not open to negotiation. A hard fork is a change of rules that makes old and new software incompatible, which is why every operator has to update before the cut-off.

A fourth term matters for understanding this case. A forward upgrade changes the rules from a future block onwards, whereas a rollback resets the chain to an earlier state and undoes transactions that have already been confirmed. Core has explicitly announced the former and ruled out the latter.

Which exchanges have stopped CORE transfers and what still works

The first publicly dated freeze comes from Coinbase. The provider's status page carries the entry "Paused Sends/Receives - Core DAO Network" with the timestamp 30 August 2026, 21:41 Pacific Time. That corresponds to 31 August 2026, 4:41 UTC. The entry still carries the status "Investigating", so it is not closed. Coinbase also writes there that buying, selling, conversions and payments in euro or dollar are not affected.

This distinction is the practically most important point of the whole episode, and it is regularly overlooked. A trading venue can keep trading running and still block the route onto the chain. You then see a normal price, you can buy and sell, but you cannot get the coins out. Anyone who only looks at the chart when checking will notice nothing of a transfer freeze.

According to consistent reporting from Cointelegraph and crypto.news, further trading venues have responded: the South Korean exchanges Bithumb and Coinone suspended deposits and withdrawals, Bitget restricted both under the heading of wallet maintenance, and LBank stopped deposits citing a request from the project. For English-speaking users, Coinbase is the most relevant of these because of how widely it is used. Which trading venues are worth considering at all, and how to recognise a regulated provider, is broken down in our crypto exchange comparison.

Our measurement of 2 September: 21 validators, then 23, then 21 again

To see what happened on the chain itself, we queried the public interface of the Core mainnet at the address rpc.coredao.org. This interface answers requests without credentials and returns, among other things, the timestamp of each block and the address that built it. Retrieving blocks over a longer period therefore shows which addresses are actively producing blocks on the network.

For this we took a sample of 600 consecutive blocks at 12:00 UTC on eight consecutive days. At a block interval of three seconds, such a sample covers half an hour, in which every active address takes its turn several times. The result is unambiguous:

Comparing the sets shows more than the bare count. The 21 addresses from 26 August and the 21 addresses from 2 September are exactly the same ones. No validators have been swapped out, in other words. Between 27 and 31 August two additional addresses joined, and those two have since disappeared from block production.

Two addresses, to the ledger: entry on 28 August, last block on 31 August

Because the samples only have daily resolution, we then narrowed down the transition. First through hourly samples of 90 blocks each across four days, then through a gapless query of the blocks around the two changes we had found. That makes it possible to name the first and last block of each address down to the individual block.

Both addresses entered block production twelve seconds apart and left it twelve seconds apart. Between those points lie three days, five hours and around 38 minutes, as well as 93,154 blocks. Since the last of those blocks we have taken 52 hourly samples of 90 blocks each, spread over the period from 31 August, 9:00 UTC, to 2 September, 13:00 UTC. Every single one of those samples showed exactly 21 addresses, and in not one of them did either of the two addresses reappear.

That establishes a number the project has not so far named: the number of validators that were additionally active in this window and then disappeared again is two.

Why no accusation follows from this coincidence in time

Caution is in order at this point, and we state it explicitly. Core has published no addresses. There is no confirmation from the project that the two addresses we measured are identical to the validators mentioned in its statement. Nor do we claim as much.

On top of that, validator sets change for entirely everyday reasons. A validator standing for election can be voted in or drop out of the vote, can cease operations, can run into technical problems, or can withdraw its own stake. An entry and a later exit are, taken on their own, a normal occurrence rather than an indication.

What can be said is the sequence in time, and that is verifiable: Coinbase paused transfers at 4:41 UTC. Around 69 minutes later, at 5:50 UTC, block production by both addresses ended. Between 5:00 and 9:00 UTC the chain fell behind its cadence. From 9:00 UTC it ran to the second again. This sequence is an observation, not an attribution of cause, and anyone can repeat it with the same queries.

The chain never stopped: 181 missing blocks and a longest pause of five seconds

Core produces blocks to a very strict cadence. Across 10,000 consecutive blocks we measured a mean interval of 3.001 seconds. At that cadence, 1,200 blocks are created per hour and 28,800 per day. That is exactly what allows a disruption to be read off without any inside knowledge.

For seven daily windows running from 12:00 to 12:00 UTC, the count produced 28,800, 28,783, 28,798, 28,798, then 28,619, followed by 28,799 and 28,792 blocks. The one clearly deviating window is the period from 30 to 31 August. We then resolved it hour by hour:

In total 181 blocks are missing, which corresponds to 543 seconds, or a little over nine minutes. The shortfall is spread across four hours and begins in the same hour in which the two addresses built their last block.

More important than the shortfall is how it came about. We counted all 4,619 block intervals in the window from 31 August, 5:00 to 9:00 UTC, individually: 4,119 intervals of three seconds, 455 of four seconds, 44 of five seconds. The longest interval in the entire window is five seconds. For comparison we counted an unremarkable window from 30 August, 20:00 to 24:00 UTC, in the same way: there, all 4,799 intervals came in at exactly three seconds.

For you as a holder, that means something very concrete. The chain did not halt, it skipped individual time slots. There was therefore no moment at which a transfer on the Core network would have been impossible. That is a marked difference from an upgrade in which a network is deliberately taken offline, as was the case most recently with the Mina hard fork with its announced network halt. The freeze was at the trading venues, not on the chain.

On the day of measurement, too, the cadence is clean. For the three hours before our final query on 2 September we counted 1,200, 1,200 and 1,200 blocks.

What cannot be seen in the fee data of the blocks

Every Core block ends with a system transaction that transfers the fees accrued in the block to a system account of the protocol. This transaction sits openly in the block and can be read out individually. We therefore checked whether the reported reward fault shows up there.

For that we drew five windows of 700 blocks each across the whole period from 28 to 31 August, 3,500 blocks in total with all their transactions, and aggregated the transferred amounts by producing address. The result: the two additional addresses sit at the bottom end. They accounted for 0.4594 CORE from 150 blocks and 0.6091 CORE from 152 blocks. The highest value for any other address was 1.7970 CORE from 155 blocks, with the middle of the field at around 0.9 CORE. The number of blocks built was likewise unremarkable and matched the share to be expected from a set working in rotation.

This is a negative result, and it belongs in an article just as much as a positive one. Nothing conspicuous can be seen in the openly visible fee flow of the blocks. The excess described by Core must therefore have arisen somewhere other than in the fees transferred per block. The distribution of the actual rewards runs through system contracts whose state can only be queried from outside for the current block. At the time of measurement, the system account to which every block transfers its fees held 467,222,693.9 CORE. How much of that belongs to which reserve cannot be determined from outside.

What a forward upgrade means and why no rollback is coming

Core has announced that it will close the fault by way of a hard fork, and has stressed that this is a forward upgrade without any reset of the network. Confirmed transactions therefore stand, and the history is not rewritten. For you that means a balance shown to you today in a self-custodied wallet does not disappear because of the fork.

What is open is the timing. Core has published neither an activation block nor a time of day, nor how many validators have to go along with it. Also open are the quantity of CORE issued in excess, the duration of the episode, and the question of whether excess tokens reached circulation. A technical report has been announced; there is no date for it.

What clean preparation for a fork looks like, when the project names a date and its requirements, is shown by our checklist for the Pasteur hard fork. The difference from the present case is precisely the missing date, and that is why waiting is the sensible posture here.

What CORE holders should check right now

The first step is to take stock, and it takes a few minutes. Look at where your CORE actually sits. On a trading venue, in a self-custodied wallet, or split across both. Everything else follows from that.

If the holding is on a trading venue, check two things separately: whether trading is running, and whether withdrawals on the Core network are possible. The second is the decisive one, and it is not in the chart but in the provider's status notice or withdrawal dialogue. If withdrawal is blocked, no detour via another chain helps as long as the provider has halted the network as a whole.

If the holding is in your own custody, it is mobile. That is exactly the difference that becomes visible on days like these: a balance in self-custody depends on the chain, while a balance on a trading venue additionally depends on that venue's decisions. Which devices are suitable for this and what sets them apart is set out in our hardware wallet comparison.

And one point that applies to every incident of this kind: in the days after such an announcement, messages pile up offering supposed compensation, an emergency migration or a check of your wallet. No hard fork and no exchange freeze ever asks for your recovery words. Anyone who asks for them wants your money. How quickly an infrastructure fault becomes a gateway for such messages was shown most recently by the compromise of an oracle across four chains.

How to check the network status and your exchange yourself in five minutes

You do not have to take our word for it, you can measure it again. The network status of Core can be queried without credentials. A call to the address rpc.coredao.org with the request eth_blockNumber returns the current block height, and a call to eth_getBlockByNumber for two blocks 1,200 numbers apart returns the hourly production via the timestamps. If the difference between the timestamps is around 3,600 seconds, the chain is running to cadence.

For the exchange side there is a shorter route. Call up your provider's status page and search for the network name. At Coinbase that is the public status page, where the entry for the Core DAO network appears with its date and processing status. Second: open the withdrawal dialogue and enter a small amount without submitting it. If the network is blocked, the dialogue will tell you so more clearly at that point than any overview.

The third step concerns your own wallet. If you hold CORE yourself and want to know whether transfers are arriving, send a small amount to a second address of your own. At a block interval of three seconds you have the answer in under a minute. We recommend the same test, incidentally, whenever withdrawal deadlines are announced, as most recently in the case of withdrawals at BitMart.

How we measured and what we could not measure

This analysis was carried out by cryptoticker.io itself on 2 September 2026. Method in one sentence: we queried the public interface of the Core mainnet on 2 September 2026 between 12:50 and 13:20 UTC, retrieved blocks individually and in batches, and narrowed the times of the changes down to the individual block by repeatedly halving the search range across the block history.

Objects examined: around 13,000 individually retrieved blocks, 3,500 of them with all contained transactions; eight daily samples of 600 blocks each; 96 hourly samples of 90 blocks each; two gapless sections of 1,400 blocks each to pin down entry and exit; one gapless section of 4,619 blocks and a comparison section of 4,800 blocks for the cadence measurement; 24 hour-boundary blocks. Every reported time of change was cross-checked against the preceding block.

Several points remain open. First: the quantity of CORE issued in excess. The public interface does not answer state queries for past blocks; for every historical state tested it reports a missing storage node. Without that access, the history of the reward accounts cannot be reconstructed from outside. Second: who stands behind the two addresses and why they ended production. Third: whether the two addresses are connected to the validators mentioned by the project; that is open and is not asserted by us. Fourth: whether and when excess tokens reached circulation. Fifth: the activation time of the hard fork, because it has not been published.

What we did not do is the following: no extrapolation to amounts we did not retrieve, no statement about the price and no statement about individual balances. There is not a single euro or dollar figure anywhere in this article, because none follows from this measurement.

Frequently asked questions on the Core DAO hard fork and the transfer freeze

Are my CORE gone because of the incident?

According to the project, the incident concerned the issuance of rewards and not user balances. The announced hard fork is a forward upgrade and does not reset confirmed transactions. A holding you hold yourself therefore stays with you.

Why can I trade but not withdraw?

Trading takes place in the provider's own books and does not need the chain. A withdrawal, by contrast, writes a transaction to the chain. If a provider blocks network access, trading remains possible while the way out is shut.

When is the hard fork coming?

No date has been published. Core has named neither an activation block nor a time of day, nor how many validators have to go along with it.

Do I have to update anything myself as a holder?

If you merely hold CORE and use it through a wallet or a trading venue, you do not have to update any network software. It is the operators of nodes and validators who have to update. Do keep your wallet app current as usual, though.

Was the chain offline at any point?

No. In our gapless count of 4,619 block intervals in the conspicuous window, the longest interval was five seconds. The chain skipped individual time slots and then returned to its cadence.

Checking the Core DAO transfer freeze: what to take away

1. Check today whether your provider has blocked the Core network, and do it in the withdrawal dialogue rather than in the chart. For as long as a freeze stands there, your holding is tradable but not movable. If you want to know in general which trading venue suits you, the criteria are in the crypto exchange comparison.
2. Decide deliberately which part of your holding should sit on a trading venue. This incident shows the difference with complete clarity: the chain carried on with a maximum interval of five seconds, while the way out at several providers stayed shut for days. Which devices are suitable for self-custody is set out in the hardware wallet comparison.
3. Wait for the technical report before you settle your view on the token. The decisive figures on the quantity issued are still missing, and without them the episode cannot be conclusively assessed. If you prefer a provider with European authorisation for cases like this, the overview of regulated crypto exchanges is the place to look.

A detailed account of the events from the perspective of several trade publications can be found at Cointelegraph, among others.

(As of September 2, 2026. This article is not investment advice. Prices and fee structures change; check the terms with the provider before you buy.)