Nov 18, 2026 · @RuthHeasman

![Key](/etchings/key.webp "right")
## Seventeen days

In the first 17 days of November 2026, attacks and failures hit systems people rely on without thinking about them. Hackers sent false nuclear alerts in Ukraine and Russia and heated supermarket freezers to 50°C in seven countries. They stole about €100 million from a French bank and planted card-stealing code on 178 organisations' websites. US air traffic, 128 Japanese banks, Microsoft's email and Chase's payments all went down.

It had the look of a Black Swan month: severe, unexpected and arriving all at once. It played out against [a war with Iran in its ninth month](https://www.iranintl.com/en/202611113106), [a strike on Saudi Arabia's East-West oil pipeline](https://aljazeera.com/news/2026/11/9/saudi-arabia-reports-new-attack-on-key-east-west-pipeline) and [an arson attempt on an Estonian drone maker](https://meduza.io/en/news/2026/11/16/estonia-blames-russian-intelligence-for-attempted-arson-at-defense-facility-announces-closure-of-all-border-crossings-with-russia-from-december-1).

This article takes nine of the month's incidents and shows where integrating Bitcoin SV (BSV) would have helped: before each one, while it unfolded, and in the aftermath.

## Before, during and after

BSV integrations help at three points in an incident.

- **Before:** signed commands and per-device keys mean a system acts only on instructions from someone authorised to give them. Wallet payments remove reusable secrets, such as card numbers, that are worth stealing.
- **During:** permissions are revocable allowances, so a compromised key is cut off everywhere at once. Integrity monitors that anchor system hashes every few seconds flag an unauthorised change as it happens. Credentials can be checked offline, and wallet-to-wallet payments keep money moving when a bank's systems stop.
- **After:** logs whose hashes are anchored on the BSV blockchain show what happened, in what order, and which key signed it. Investigators can check a stated cause against the record instead of taking it on trust. The same record can show that an outage was not a hack.

![before-during-after](/api/storage/objects/uploads/fded4f20-3567-462f-8b36-78afc4375fcf)


The aftermath step rests on peer-reviewed work: Haber and Stornetta's linked timestamping (*Journal of Cryptology*, 1991) and Crosby and Wallach's tamper-evident logging (USENIX Security, 2009). BSV adds a public place to anchor, at fees low enough to anchor every few minutes.

## Integrity monitoring: flag the hack, or prove there wasn't one

One BSV integration already in production works at both ends of an incident. [SmartLedger](https://smartledger.technology/)'s Sentinel Node, the first module of its Certihash suite, built with IBM Consulting, takes a snapshot of a system every 10 seconds and stores its hash on the BSV blockchain ([BSV Association](https://bsvassociation.org/?p=2335)). A change that does not match an authorised one raises an alert ([BSV Blockchain](https://bsvblockchain.org/?p=15187)).

Each snapshot can be issued as a certificate under NotaryHash, [SmartLedger](https://smartledger.technology/)'s BRC-220 standard: a signed hash anchored on-chain, which anyone can verify offline and against block headers alone, without the notary ever seeing the original data ([BRC-220](https://hub.bsvblockchain.org/brc/apps/0220)).

That gives it two uses in November's incidents:

- **Flag an attack as it happens.** Where attackers changed settings or code (the freezers, the Port of Santos, Optimizely, and the Volt Typhoon and Salt Typhoon campaigns), the first unauthorised change triggers an alert within seconds.
- **Prove an outage was not an attack.** Where the cause was a failed update or faulty equipment (the FAA, Microsoft, the shinkin banks and Chase), an unbroken run of snapshots shows that nothing changed apart from the approved update. "No sign of a cyberattack" becomes a statement anyone holding the record can check.

The proof covers the systems being monitored, so the monitor needs to watch every system in the path of a failure.

The records also have to survive. An anchor proves a log unchanged only while the log itself still exists. Proof of Keep, the author's storage-proof scheme, pays storage operators only when they prove through on-chain challenges that they still hold the data, so logs stay retrievable for as long as investigators need them ([paper](https://permastackpress.com/ruth-heasman/proof-of-keep), [code](https://github.com/Ruthheasman/proof-of-keep)).

## The incidents at a glance

Nine incidents between 1 and 17 November, and what BSV would have changed at each stage.

| Date (2026) | Incident | Before | During | After |
| --- | --- | --- | --- | --- |
| 1–2 Nov | Air-raid alert apps hacked | Apps display only alerts signed by issuing agencies | A stolen key is revoked for every app at once | Public record of every alert issued |
| 1–2 Nov | Supermarket freezers set to 50°C | Each controller has its own key; commands must be signed | Unauthorised change flagged in seconds; operator key cut off fleet-wide | Signed setpoint log shows who changed what |
| 4 Nov | Crédit Mutuel Arkéa robbed of about €100m | Provider acts only inside a signed spend policy; miners enforce the payee list | Policy frozen; circuit breakers stop abnormal transfers | Anchored instruction log for tracing funds |
| 6 Nov | 128 shinkin banks offline | Network changes anchored before rollout | Wallet-to-wallet payments keep working | Change log points to the cause |
| 6–10 Nov | Port of Santos systems down | Gate passes and releases issued as signed tokens | Gates check passes offline; file changes flagged early | Access log shows who entered and collected what |
| 12 Nov | Microsoft 365 outage | Network changes anchored when made | Engineers work from a complete change history | Customers check the stated cause against the record |
| 13 Nov | FAA nationwide ground stop | AI tool and network changes logged and anchored | Log rules the AI tool in or out within minutes | AI fault and hacking ruled in or out from the record |
| 16 Nov | Chase and Zelle delays | Customers hold wallets as a second payment rail | Wallet-to-wallet payments keep working | Change log shows which change caused the bug |
| 16 Nov | Optimizely scripts hijacked | Only owner-signed releases load; no card numbers to skim | Injected code flagged in seconds and revoked | Release log shows when exposure began and ended |

## The incidents, day by day

### Air-raid alert apps, 1–2 November

Late on 1 November, hackers took over Ukraine's Air Alert app and sent false warnings of nuclear and chemical attacks and street fighting in Kyiv ([Kyiv Independent](https://kyivindependent.com/ukraines-air-raid-alert-system-hacked-air-force-says-as-false-nuclear-chemical-warnings-sound/)). Hours later, Russia's Emergency Situations Ministry said its own warning app had been hacked ([The Record](https://therecord.media/ukraine-hackers-russia-alert-system)).

**Before.** Every alert carries the issuing agency's signature, and the app displays only signed alerts. Nuclear and chemical alerts need signatures from two of three separate agencies, so taking over one server is not enough to issue one.

**During.** If a signing key is stolen, its revocation is published on-chain and every app stops trusting it at once.

**After.** The hash of every alert is anchored on BSV, so anyone can check which alerts the authority actually issued, and when. The same design suits the US Emergency Alert System, which ran its nationwide test on 17 November ([FCC](https://docs.fcc.gov/public/attachments/DA-26-989A1.txt)).

A false alert gets a signed retraction, anchored like any other record, and the retraction follows every record built on it. In Ukraine that included a volunteer-run alert map that displayed the false warnings ([Kyiv Independent](https://kyivindependent.com/ukraines-air-raid-alert-system-hacked-air-force-says-as-false-nuclear-chemical-warnings-sound/)). The [HelpPeer provenance notary](https://x.com/ruthheasman/status/2094367525224034804), built by the author, implements this: retractions are reversible, and a "taint" flag passes to everything derived from the retracted record.

### Supermarket and military freezers, 1–5 November

Overnight on 1–2 November, attackers set refrigeration controllers to 50°C in supermarkets across seven countries, forcing stores to throw away food ([Reuters via Taiwan News](https://www.taiwannews.com.tw/en/news/6471142), [IBTimes](https://www.ibtimes.co.uk/europe-supermarket-refrigeration-cyberattacks-food-waste-1829795)). On 5 November, CISA, the FBI and three other US agencies warned that attackers were pushing units into continuous heating and defrost cycles, after freezer failures at US military commissaries ([BleepingComputer](https://bleepingcomputer.com/news/security/fbi-cisa-warn-of-cyberattacks-targeting-refrigeration-systems), [Nextgov](https://www.nextgov.com/cybersecurity/2026/11/us-agencies-warn-refrigeration-hacks-amid-suspected-attacks-military-bases/417071/)).

**Before.** Each controller gets its own key, derived per device and operator under BRC-42, replacing shared or default passwords. A setpoint change runs only when an authorised operator key signs it, and a change outside the safe range needs a second signature.

**During.** Operator rights are revocable allowances, the model set out in [*The Payment Is the Permission*](https://x.com/ruthheasman/status/2102790651490242994). A compromised key is cut off across every store at once.

An integrity monitor on the controllers flags the change the moment it is made. At one Stockholm store the attack happened at 2 am but was noticed only hours later ([Kuwait Times, citing Reuters](https://kuwaittimes.com/article/52864/business/cyberattack-on-fridges-hits-supermarkets-across-europe/amp)).

**After.** The signed setpoint log, anchored every few minutes, shows which key changed which unit and when. Because nobody can edit it afterwards, insurers and investigators can rely on it.

### Crédit Mutuel Arkéa, 4 November

Attackers used one of its service providers' platforms to reach the French bank and moved about €100 million to other banks. Some of it was converted into cryptocurrency ([GovInfoSecurity](https://www.govinfosecurity.com/french-bank-loses-100-million-euros-in-cyberattack-a-33309), [Arab News](https://www.arabnews.pk/world/french-bank-says-cyberattack-stole-around-100m-3006123)).

**Before.** The provider acts only inside a spend policy the bank has signed. BRC-181, the Wallet-Enforced Autonomous-Agent Spend Policy, already defines this for automated spenders: approved payees, caps per payment, per period and in total, and an instant freeze ([BRC-181](https://hub.bsvblockchain.org/brc/wallet/0181), [CoinGeek](https://coingeek.com/new-bsv-standard-lets-humans-control-ai-agents-spending/)). A second, miner-enforced layer comes from the author's draft Mandala Covenant Profile: tokenised value can move only to a committed payee list and within a per-transaction limit unless a guardian key signs. A stolen key can make only allowed payments until the guardian acts. The profile builds on BRC-162 Mandala tokens, so it applies where deposits are tokenised ([BRC-162](https://hub.bsvblockchain.org/brc/tokens/0162)).

**During.** The bank freezes the provider's policy instantly, and BRC-181's circuit breakers stop activity that looks abnormal until a person clears it. As a fallback, large outbound transfers can wait in an output that a bank recovery key can reclaim. On BSV that window needs a covenant script that reads the transaction's lock time, and the design needs technical review before use.

**After.** An anchored log of every instruction the provider sent gives investigators an exact sequence to trace the funds.

### 128 shinkin banks, 6 November

A failure in the shared system behind 128 shinkin banks in western Japan stopped ATMs, branch counters and online banking across 3,430 branches. That network had been updated in mid-October ([Mix Vale](https://www.mixvale.com.br/2026/11/06/128-shinkin-banks-in-western-japan-lose-atms-and-counters-after/), [Outlook India](https://www.outlookindia.com/international/japan-banking-outage-128-shinkin-banks-western-japan)).

**Before.** Every change to the shared network is anchored before rollout, with its approval and a hash of the configuration.

**During.** Customers and shops with BSV wallets keep paying each other directly. A payment verified by SPV does not pass through any bank's systems.

**After.** The anchored change log shows what changed and when, so regulators and member banks can check the cause for themselves. Integrity snapshots back Shinkin Central Bank's statement that no cyberattack was detected with a record others can check.

### Port of Santos, 6–10 November

From 6 November, the port authority's identification, access-control and gate systems at Latin America's largest port were down. On 10 November the authority confirmed a cyberattack ([FreshPlaza](https://www.freshplaza.com/article/9891615/system-outage-disrupts-operations-at-brazil-s-port-of-santos/), [Baguete](https://www.baguete.com.br/noticias/porto-de-santos-sofre-ataque-cibernetico)).

**Before.** Gate passes and container releases are issued as signed credentials. A container release is a one-time token held by the authorised haulier, so it cannot be copied.

**During.** Gate readers verify passes offline against the issuer's key, so trucks keep moving while central systems are down.

The attackers reportedly demanded payment to restore services ([FreshPlaza](https://www.freshplaza.com/article/9891615/system-outage-disrupts-operations-at-brazil-s-port-of-santos/)). An integrity monitor on the authority's servers flags mass file changes the moment they start, so affected machines can be isolated early.

Recovery depends on backups. By 12 November the authority had recovered 70% of its backed-up files ([Times Brasil](https://timesbrasil.com.br/empresas-e-negocios/tecnologia-e-inovacao/porto-de-santos-recupera-70-dos-backups-apos-ataque-hacker/)). Under Proof of Keep, backup operators are paid only when they prove through regular on-chain challenges that they still hold the data, so the authority would know its backups were whole before it needed them ([paper](https://permastackpress.com/ruth-heasman/proof-of-keep), [code](https://github.com/Ruthheasman/proof-of-keep)).

**After.** An anchored access log shows who entered, who collected which container, and when.

### Microsoft 365, 12 November

Outlook, Teams and Copilot went down for users worldwide. Microsoft traced the outage to a problem from an earlier network-capacity upgrade that surfaced after a newer change ([BleepingComputer](https://bleepingcomputer.com/news/microsoft/massive-microsoft-365-outage-disrupts-outlook-teams-copilot-worldwide), [Latin Times](https://www.latintimes.com/microsoft-365-global-outage-outlook-teams-copilot-601369)).

**Before.** Every network change is anchored when it is made, with its approval and a hash of the configuration before and after.

**During.** Engineers work from a complete, ordered change history to find the faulty change.

**After.** Customers can check Microsoft's explanation against the anchored record. Integrity snapshots also show whether anything changed outside the approved upgrade, which backs Microsoft's statement that the outage was not malicious.

### FAA ground stop, 13 November

On 10 November the FAA switched on SMART, an $875 million AI route advisor, at all 22 of its en route centres ([Beinsure](https://beinsure.com/news/faa-smart-system-ai-rollout-and-nationwide-flight-disruption/)). Three days later, a telecom outage at those centres held every US departure for about four hours ([AeroTime](https://www.aerotime.aero/articles/faa-nationwide-ground-stop-atc-communications-november-2026/amp)). SMART's developer said its system was not the cause; the Homeland Security Secretary said nothing could be ruled out ([KPTV](https://www.kptv.com/2026/11/16/questions-swirl-around-faas-new-smart-ai-system-though-officials-say-no-link-ground-stop/?outputType=amp), [TheWrap](https://thewrap.com/media-platforms/politics/markwayne-mullin-ground-stop-isnt-another-september-11-cnn)).

**Before.** Every SMART recommendation is logged with hashes of its inputs, the model version and whether a human accepted it. The log is signed by the tool's own key and anchored every minute, and every change to the telecom network is anchored the same way. This is the audit trail set out in *The Payment Is the Permission*.

**During.** Within minutes, the log shows whether any SMART output was in play when the network failed, so the response can focus on the actual cause.

**After.** "Was the AI involved?" is answered from the record, and anyone given the log can check it against the anchors.

Integrity snapshots of the telecom network and of SMART itself show whether anything changed apart from the approved update. That turns the Homeland Security Secretary's "98 percent" certainty that no bad actor was involved into a record others can check ([Daily Beast](https://www.thedailybeast.com/trump-goon-markwayne-mullin-serves-up-dizzying-word-salad-on-flight-chaos)).

### Chase and Zelle, 16 November

A software bug at JPMorgan Chase slowed transfers and Zelle payments for its customers and for banks transacting with it ([AP via ABC11](https://abc11.com/post/customers-nations-largest-banks-are-reporting-difficulty-accessing-accounts-online/20045188/), [WebProNews](https://www.webpronews.com/outages-impact-chase-bank-of-america-wells-fargo-us-bank-and-more/)).

**Before.** Customers and merchants hold BSV wallets as a second payment rail.

**During.** Wallet-to-wallet payments keep working because they do not touch any bank's core systems. The recipient confirms each payment against block headers and a Merkle proof.

**After.** Chase's anchored change log shows which change introduced the bug, and integrity snapshots back its statement that the problem was not cyber-related.

### Optimizely, 16 November

For about 16 hours, scripts served by the testing platform Optimizely carried a card skimmer and a crypto clipper onto 178 organisations' websites, including CNN, the BBC, Zoom and Salesforce. Optimizely says an unauthorised party exploited a flaw in its account signup flow ([StepSecurity](https://www.stepsecurity.io/blog/optimjack-optimizely-hosted-scripts-crypto-clipper-card-skimmer)).

**Before.** Each customer signs its own script releases, and each release's hash is anchored. Pages load only code that matches the owner's signed release log, so a new account cannot publish into another customer's project. A BRC-100 wallet payment is a signature over one transaction to one merchant-derived key, so there is no card number to skim. Paying a verified counterparty at an address derived under BRC-42 removes the copied address a clipper swaps.

**During.** When a bad release is spotted, its owner revokes it and every site stops loading it.

An integrity monitor watching the hosted snippets flags the injected loader within seconds of it appearing, instead of after about 16 hours. Customers can run the same check on the third-party scripts their own pages load.

**After.** The anchored release log shows exactly when the bad code appeared and when it was revoked, so each organisation knows which visitors to warn.

Two features of the HelpPeer notary fit here. Its observation mode lets each customer record the vendor's script exactly as served, then re-fetch and compare it: identical, different or unavailable. Its taint model marks every page that served a retracted release, so each affected site can see it was exposed.

## Already inside: three campaigns running before November

The same integrations help against three campaigns that were under way before the month began.

- **Water utilities.** Hackers reached control systems at water utilities in at least 12 US states in July and August, and the FBI told utilities to prepare to run by hand ([Axios](https://www.axios.com/2026/08/04/water-cyberattacks-us-iran)). Signed setpoint commands and revocable operator keys, as for the freezers, close the same gap.
- **Volt Typhoon.** The Chinese state group built a new hacking network after a US crackdown, with Energy Department facilities among its targets ([Nextgov](https://www.nextgov.com/cybersecurity/2026/10/chinas-volt-typhoon-quietly-built-another-hacking-network-following-us-crackdown-new-findings-say/416848/)). Its intruders hide by editing logs; log hashes anchored on BSV every few minutes make any rewrite visible, and an integrity monitor flags the new tools and configuration changes they leave behind.
- **Salt Typhoon.** The group reached the lawful-intercept systems US carriers run for wiretaps, and the FBI said in February 2026 that it was still active ([CyberScoop](https://cyberscoop.com/fbi-salt-typhoon-ongoing-threat-cybertalks-2026/), [DeepStrike](https://deepstrike.io/blog/salt-typhoon)). If every wiretap activation carries a signed warrant token with its hash anchored, a tap with no matching record flags an intruder. Integrity snapshots of router configurations flag changes as they are made.

## The building blocks

Every integration in this article uses a published standard, a product in service, or open work by the author.

| Building block | What it does | Where it helps | Link |
| --- | --- | --- | --- |
| BRC-181, Wallet-Enforced Autonomous-Agent Spend Policy | Wallet obeys only a signed policy: approved payees, spending caps, instant freeze, circuit breakers | Crédit Mutuel Arkéa, freezer operators, AI tools | [BRC-181](https://hub.bsvblockchain.org/brc/wallet/0181) |
| *The Payment Is the Permission* (the author, 2026) | Identity, an audit trail and instant revocation for AI agents and other automated actors | FAA SMART, freezer operators | [X article](https://x.com/ruthheasman/status/2102790651490242994), 23 September 2026 |
| Mandala Covenant Profile (the author's draft BRC) | Miners refuse transfers outside a committed payee list or above a per-transaction limit | Crédit Mutuel Arkéa | Builds on [BRC-162](https://hub.bsvblockchain.org/brc/tokens/0162) |
| Sentinel Node ([SmartLedger](https://smartledger.technology/), Certihash) | System snapshots every 10 seconds, hashes on BSV, alerts on unauthorised change | Freezers, Santos, Optimizely, FAA, Microsoft, shinkin banks, Chase | [BSV Association](https://bsvassociation.org/?p=2335) |
| NotaryHash, BRC-220 (SmartLedger) | Signed-hash certificates verifiable offline and against block headers | Every "after" record | [BRC-220](https://hub.bsvblockchain.org/brc/apps/0220) |
| HelpPeer provenance notary (the author) | Batch-anchored records, reversible retraction, taint that follows derived records, read-only observation of other systems | Air-raid alerts, Optimizely | [X article](https://x.com/ruthheasman/status/2094367525224034804) |
| Proof of Keep (the author) | Storage operators paid only when they prove they still hold the data | Santos backups; keeping logs retrievable | [Paper](https://permastackpress.com/ruth-heasman/proof-of-keep), [code](https://github.com/Ruthheasman/proof-of-keep) |
| @bsv/sdk (BSV Association) | SPV verification and BRC-100 wallets | Chase and shinkin fallback payments | [npm](https://www.npmjs.com/package/@bsv/sdk) |

## What it adds up to

Across the nine incidents, four integrations do most of the work:

1. **Signatures instead of reusable secrets:** card numbers, default passwords and shared operator logins.
2. **Authority that travels with every command:** signed, scoped and revocable allowances for operators, vendors and AI tools.
3. **Logs anchored before anything goes wrong:** change records, AI actions and alert issuance.
4. **Integrity snapshots anchored every few seconds:** they flag an attack as it happens, and show when an outage was not one.

The building blocks are on BSV mainnet now. The remaining work is integration.

[Watch the Explainer Video](https://youtu.be/zQJyrtgQO2Y?is=lk_3Iu7ojQG0PiwK)

## Sources

Incident reporting is linked in each section. Background references, cited from memory and not linked:

- Taleb, N. N. (2007). *The Black Swan*.
- Haber, S. and Stornetta, W. S. (1991). How to time-stamp a digital document. *Journal of Cryptology* 3(2).
- Crosby, S. A. and Wallach, D. S. (2009). Efficient data structures for tamper-evident logging. *USENIX Security Symposium*.