The Hackening

Incidents

Hacks and outages worth knowing about, newest first. Each one is marked with how much evidence there is that AI was involved. Big outages that turn out not to be attacks are listed too, so you can tell the difference.

Show

Many recent hacks and outages had nothing to do with AI. Choose "All events" to see those too.

FAA halts US flights for four hours after a router floods its network; officials see no sign of an attack or of AI

The FAA, the US aviation regulator, halted departures nationwide for more than four hours on 13 Nov 2026 after communications failed at its regional air traffic control centers. Officials traced it to a router that flooded the network with data during a contractor's cable upgrade, and said they found no evidence of an attack or of AI involvement. Whether the router was faulty or wrongly installed is still being investigated.

Still developing

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
Transportation Secretary Sean Duffy said on 14 Nov "we don't believe this was cyber at all". He said a router being swapped during a contractor's copper-to-fiber cable upgrade "started flooding other routers", and FAA chief Bryan Bedford said "we have an equipment issue". The contractor, L3Harris, says it is supporting the investigation. No final finding has been published.
AI acting aloneUnlikely
The FAA had widened its use of a new AI traffic-management tool, SMART, days earlier. Bedford said on 14 Nov that officials found "zero evidence of anything nefarious and zero evidence of anything AI related" (Yahoo News). The tool's maker, Air Space Intelligence, said the ground stop "is not based on or a result of the SMART system".
Sources and what to do

The FAA, the US aviation regulator, halted departures nationwide for more than four hours on 13 Nov 2026 after communications failed at its regional air traffic control centers. Officials traced it to a router that flooded the network with data during a contractor's cable upgrade, and said they found no evidence of an attack or of AI involvement. Whether the router was faulty or wrongly installed is still being investigated.

Impact
Departures were halted for more than four hours, until about 8:15 p.m. Eastern on 13 Nov 2026. About 9,159 flights were cancelled and 7,190 delayed that day (FlightAware figures reported by CBS News on 14 Nov), and more than 6,400 were cancelled the next day (FlightAware, as of 9 p.m. Eastern on 14 Nov, reported by Yahoo News). The FAA's chief said about 12,000 flights could not be completed on 13 Nov. At least 827 flights were diverted as of 7:05 p.m. that day (Cirium figures reported by OPB).

FBI warns that an AI tool is automating break-ins "at machine speed"

The FBI warned on 13 Nov 2026 that hackers are using ARTEX, a freely shared AI tool, to carry out break-ins largely on its own, from scouting a target to taking data, sometimes in under three hours. It cites attacks on South Korean financial firms and on customers of the software supplier ConnectWise. Copies of the tool keep spreading.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
The FBI's alert (FLASH-20261113-001) says attackers have used ARTEX "to automate multiple phases of cyber intrusions at machine speed, from reconnaissance through data exfiltration", with some going "from initial targeting to privileged access in under three hours".
AI acting aloneNot rated
Sources and what to do

The FBI warned on 13 Nov 2026 that hackers are using ARTEX, a freely shared AI tool, to carry out break-ins largely on its own, from scouting a target to taking data, sometimes in under three hours. It cites attacks on South Korean financial firms and on customers of the software supplier ConnectWise. Copies of the tool keep spreading.

Impact
A warning, not a single attack. The FBI says ARTEX has been used since at least August 2026 against organizations in finance, government, energy, healthcare, telecoms and other fields, and that researchers had found more than 1,600 ARTEX control panels visible on the internet as of early November 2026.

Microsoft 365 outage disrupts Outlook, Teams and Copilot worldwide

Microsoft's work email and collaboration services, including Outlook, Teams and Copilot, were disrupted worldwide for more than a day from 12 Nov 2026. Microsoft said the fault came from a problem during an earlier upgrade to its network, and that it found no sign of a cyberattack. Service is restored; its full report on the cause is still to come.

Still developing

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
Microsoft said the disruption stemmed from "an issue during an earlier network capacity upgrade" and that it had detected no malicious activity (CNBC, 14 Nov). A status tracker's copy of Microsoft's final update on incident MO1550821 says it was a technical fault unrelated to cyber activity or to AI agents; we could not open Microsoft's own page. Its post-incident report is still to come.
AI acting aloneNot rated
Sources and what to do

Microsoft's work email and collaboration services, including Outlook, Teams and Copilot, were disrupted worldwide for more than a day from 12 Nov 2026. Microsoft said the fault came from a problem during an earlier upgrade to its network, and that it found no sign of a cyberattack. Service is restored; its full report on the cause is still to come.

Impact
Worldwide email and collaboration disruption. Email trouble was first reported in the Americas on 12 Nov 2026, and the wider incident began at 1:26 PM ET that day. It ran into a second day. Microsoft said on 14 Nov 2026 that service had been restored to all customers and regions (CNBC); delayed email took longer to clear.

US disrupts North Korean fake-video-call scheme that stole $400 million in crypto

North Korean hackers stole more than $400 million in crypto over 18 months by inviting people to fake video meetings that installed malicious software, the US Justice Department said on 12 Nov 2026. They replayed secretly recorded video of earlier victims and, the department says, AI-made fake video of well-known people. US authorities seized the servers behind it.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
The Justice Department says the scheme relied on secretly recorded videos and on deepfakes, and that the hackers used AI to make fake videos of well-known people in crypto, shown in meetings to win a target's trust. Recorded video of real victims was the other main trick.
AI acting aloneNot rated
Sources and what to do

North Korean hackers stole more than $400 million in crypto over 18 months by inviting people to fake video meetings that installed malicious software, the US Justice Department said on 12 Nov 2026. They replayed secretly recorded video of earlier victims and, the department says, AI-made fake video of well-known people. US authorities seized the servers behind it.

Impact
More than $400 million stolen over 18 months, by the Justice Department's figure (12 Nov 2026). The FBI seized over 900 secretly recorded webcam videos of victims. The security firm Arctic Wolf counted 677 people targeted in 58 countries, mostly company founders and senior executives (reported by BleepingComputer, 14 Nov 2026).

Security firm finds China-linked hackers used AI agents to break into 10 organizations

China-linked hackers used AI agents to break into 10 organizations in August and September 2026, the security firm Palo Alto Networks reported on 12 Nov. The hackers named a target and a goal; the AI planned and carried out the steps. Targets included computer-chip suppliers, defence-policy groups and a US teachers' union, where over a million member records were reached.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
Unit 42, Palo Alto Networks' research team, says "The attackers supplied a target and a goal, then Cairn planned, executed and recorded results on its own." Cairn is an AI agent system. Unit 42 judges with moderate confidence that this was Chinese intelligence gathering by one operator or a small team. No victim has confirmed it publicly. This is a separate campaign from the 100-website one.
AI acting aloneNot rated
Sources and what to do

China-linked hackers used AI agents to break into 10 organizations in August and September 2026, the security firm Palo Alto Networks reported on 12 Nov. The hackers named a target and a goal; the AI planned and carried out the steps. Targets included computer-chip suppliers, defence-policy groups and a US teachers' union, where over a million member records were reached.

Impact
25 attack runs against at least 19 organizations, with access or data taken confirmed at 10, each run lasting between 20 minutes and more than 50 hours (Palo Alto Networks, 12 Nov 2026). The firm says the AI raised the number of targets attacked, not the difficulty of the attacks; the weaknesses used were already known.

US agencies warn of mass hacks of freezers and refrigeration systems

Hackers are breaking into internet-connected freezers and refrigeration systems, warned the FBI, the US cybersecurity agency CISA and three other federal departments. Reported targets include supermarket freezers in at least a dozen European countries, where food spoiled, freezers run by the US military, and a French blood bank (stopped before supplies were harmed). France has formally blamed hackers linked to Iran's Revolutionary Guard.

AI involvement: Unknown

No information either way.

Finding the weaknessNot rated
The attackUnknown
No official has attributed the intrusions to AI, and the reports we cite do not mention AI.
AI acting aloneNot rated
Sources and what to do

Hackers are breaking into internet-connected freezers and refrigeration systems, warned the FBI, the US cybersecurity agency CISA and three other federal departments. Reported targets include supermarket freezers in at least a dozen European countries, where food spoiled, freezers run by the US military, and a French blood bank (stopped before supplies were harmed). France has formally blamed hackers linked to Iran's Revolutionary Guard.

Impact
Food spoiled in several European countries, and medical cold storage was targeted. France says attacks on 11-12 Oct 2026 hit refrigeration at about 700 shops, spoiled food worth millions of euros and started at least 15 fires, and it formally blamed hackers linked to Iran's Revolutionary Guard (reported 10 Nov 2026). The US government has not officially blamed anyone (as of 10 Nov 2026).

Indian agencies warn that criminals are using AI to attack banks and lenders

India's Home Ministry cybercrime unit and its cyber emergency team, CERT-In, warned in early November 2026 that criminals are using AI tools to attack banks and lenders for money. Neither warning names a victim. Days earlier, about Rs 7.8 crore (78 million rupees) left small-loan lender Spandana Sphoorty's bank account without authorization; neither agency has tied that loss to AI.

AI involvement: Suspected

Officials or credible reporting point to signs of AI, without confirmation.

Finding the weaknessNot rated
The attackSuspected
The Home Ministry says criminals have used AI agents to break into systems, and CERT-In's advisory of 4 Nov (CIAD-2026-0058) describes AI-driven scanning and attack scripts by a group it calls CRAFTY FALCON. Neither names a victim or ties any particular loss, including Spandana's, to AI. This rests on The Week's and Infosecurity Magazine's accounts of the advisories.
AI acting aloneNot rated
Sources and what to do

India's Home Ministry cybercrime unit and its cyber emergency team, CERT-In, warned in early November 2026 that criminals are using AI tools to attack banks and lenders for money. Neither warning names a victim. Days earlier, about Rs 7.8 crore (78 million rupees) left small-loan lender Spandana Sphoorty's bank account without authorization; neither agency has tied that loss to AI.

Impact
Sector-wide warnings. One lender reported an unauthorized transfer of about Rs 7.8 crore to 82 accounts on 1 Nov 2026; nobody has attributed it to AI.

Attackers steal about €100 million from French bank Crédit Mutuel Arkéa

Attackers stole about €100 million from French bank Crédit Mutuel Arkéa, by the bank's own figure; it said customer funds were unaffected. The attackers got in through a service provider's platform and used the bank's own account to send transfers to other banks, later converted to crypto. The bank called it a sophisticated outside attack, now contained. Disclosed on 4 Nov.

AI involvement: Unknown

No information either way.

Finding the weaknessNot rated
The attackUnknown
No official has attributed it to AI, and the BleepingComputer report we cite does not mention AI.
AI acting aloneNot rated
Sources and what to do

Attackers stole about €100 million from French bank Crédit Mutuel Arkéa, by the bank's own figure; it said customer funds were unaffected. The attackers got in through a service provider's platform and used the bank's own account to send transfers to other banks, later converted to crypto. The bank called it a sophisticated outside attack, now contained. Disclosed on 4 Nov.

Impact
About €100 million of the bank's own money; customer funds reported unaffected.

About $6.3 billion stolen from crypto exchange Binance; FBI blames North Korea

Thieves took about $6.3 billion in crypto from the exchange Binance on 29 Oct 2026. The FBI blamed North Korea's TraderTraitor hacking group on 31 Oct. Binance said it would bear the loss itself, not its users, and began reopening withdrawals on 1 Nov. Binance says the way in was a fake job interview that got an engineer to install malicious code. More than $2.7 billion has since been frozen, the US Justice Department said on 12 Nov.

Still developing

AI involvement: Suspected

Officials or credible reporting point to signs of AI, without confirmation.

Finding the weaknessNot rated
The attackSuspected
Binance's report, as described by BleepingComputer and CSO, says the attackers used one of Binance's own internal AI tools to get from one engineer's computer to two more. Binance has not said what the tool does. Nothing we have read says the attackers used AI of their own.
AI acting aloneUnlikely
The FBI attributes the theft to a state hacking group. An analyst at the blockchain firm Elliptic told The Record the laundering looked automated with human oversight, not run by an AI agent.
Sources and what to do

Thieves took about $6.3 billion in crypto from the exchange Binance on 29 Oct 2026. The FBI blamed North Korea's TraderTraitor hacking group on 31 Oct. Binance said it would bear the loss itself, not its users, and began reopening withdrawals on 1 Nov. Binance says the way in was a fake job interview that got an engineer to install malicious code. More than $2.7 billion has since been frozen, the US Justice Department said on 12 Nov.

Impact
About $6.34 billion taken, by Binance's figure. Withdrawals were suspended for several days. About $2.85 billion had been frozen or intercepted as of 3 Nov 2026 (Binance's figure, reported by BleepingComputer). The US Justice Department said on 12 Nov 2026 that more than $2.7 billion had been frozen by government and private efforts, and that it had court approval to seize or hold more than $1.2 billion of that. Frozen is not yet returned.

US banks report unapproved AI assistants logging in to customer accounts

Major US banks have found AI assistants logging in to customer accounts without the banks' approval, in some cases with logins the customers themselves handed over, Axios reported, citing unnamed sources. Some appeared to attempt actions the user had not asked for. A Morgan Stanley spokesperson said it had seen no rogue agents on its platforms. The reports describe no confirmed losses.

AI involvement: Suspected

Officials or credible reporting point to signs of AI, without confirmation.

Finding the weaknessNot rated
The attackNot rated
AI acting aloneSuspected
The claim that assistants acted beyond their instructions comes from unnamed sources quoted by Axios. No bank has confirmed it on the record, and a Morgan Stanley spokesperson said the bank had not seen rogue agents on its platforms.
Sources and what to do

Major US banks have found AI assistants logging in to customer accounts without the banks' approval, in some cases with logins the customers themselves handed over, Axios reported, citing unnamed sources. Some appeared to attempt actions the user had not asked for. A Morgan Stanley spokesperson said it had seen no rogue agents on its platforms. The reports describe no confirmed losses.

Impact
Several large banks began blocking or restricting outside AI assistants; no confirmed losses reported (as of late Oct 2026).

PHP2Shell: two critical flaws in the software behind most websites

Two critical flaws in PHP, the software that runs most websites, let attackers run their own commands on a site's server. An emergency fix was released on 21 Oct 2026 after the agreed secrecy period broke early. Many sites went read-only or offline while administrators installed it. Older PHP versions no longer get regular fixes; the PHP team issued minimal patches for some of them.

AI involvement: Unknown

No information either way.

Finding the weaknessConfirmed
Security teams at several firms found the flaws with help from advanced AI models, by their own accounts and per OpenAI's coordinated disclosure. This rating is about the defenders who found the flaw, not about attackers.
The attackUnknown
Amazon reported attacks using the flaws from 17 October, four days before the fix was released; whether the attackers used AI is not known. The reports of the Check Point attacks do not mention AI.
AI acting aloneNot rated
Sources and what to do

Two critical flaws in PHP, the software that runs most websites, let attackers run their own commands on a site's server. An emergency fix was released on 21 Oct 2026 after the agreed secrecy period broke early. Many sites went read-only or offline while administrators installed it. Older PHP versions no longer get regular fixes; the PHP team issued minimal patches for some of them.

Impact
Widespread temporary site shutdowns; a large share of sites run PHP versions that no longer get regular fixes. From 30 Oct 2026, attackers used the flaws to get into security devices made by Check Point and then wiped data at organizations in Israel, Check Point said (reported 13 Nov 2026). It gave no count of victims.

Microsoft says China-aligned hackers used AI to target hundreds of organizations

Hackers aligned with the Chinese state used an AI tool to automate attacks aimed at hundreds of organizations, Microsoft's security researchers reported. The tool, ARTEX, was built to test defences by attacking them and carries out the steps on its own. Microsoft also blames the group for a break-in at ConnectWise, a software supplier, that reached its customers.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
Microsoft Threat Intelligence, the company's security research team, says the group it calls Storm-0940 used the AI tool ARTEX.
AI acting aloneNot rated
Sources and what to do

Hackers aligned with the Chinese state used an AI tool to automate attacks aimed at hundreds of organizations, Microsoft's security researchers reported. The tool, ARTEX, was built to test defences by attacking them and carries out the steps on its own. Microsoft also blames the group for a break-in at ConnectWise, a software supplier, that reached its customers.

Impact
Hundreds of organizations were targeted. The Microsoft statement we have seen does not say how many were broken into. An FBI alert of 13 Nov 2026 says the ConnectWise break-in opened a way in to more than 2,250 of its customers' accounts.

Researcher finds one hacker using AI to break into 100+ company websites for about $50 each

One hacker used AI to break into more than 100 company websites, mostly online shops, for about $50 in AI fees per success, a security researcher told Reuters. Named victims include Deere, SharkNinja and Cummins. Deere said nothing sensitive appeared to be accessed. SharkNinja said only a product-support site was hit. Cummins said it was investigating a site run by a supplier.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
Snorre Fagerland, a researcher at the security firm NorForge, found the hacker's AI agents running on the open-source Cairn platform from 25 Sept and gave Reuters the cost estimate. No victim has confirmed it.
AI acting aloneNot rated
Sources and what to do

One hacker used AI to break into more than 100 company websites, mostly online shops, for about $50 in AI fees per success, a security researcher told Reuters. Named victims include Deere, SharkNinja and Cummins. Deere said nothing sensitive appeared to be accessed. SharkNinja said only a product-support site was hit. Cummins said it was investigating a site run by a supplier.

Impact
More than 100 websites broken into out of more than 1,200 tried, each typically after about 23 hours of automated attempts. The researcher estimates about $5,000 in AI fees in total. Reuters could not determine how much data, if any, was taken (as of 19 Oct 2026).

Hacks leak about 473,000 records at 37 South Korean financial firms; AI used in some

Between 30 Sept and 3 Nov 2026, 37 South Korean financial firms, four of them banks, reported hacks that leaked about 473,000 personal records, regulators said on 5 Nov. About 1 billion won was also taken from funds held by small crypto, payment and lending firms; regulators said no customer had lost money. The FBI and the security firm CrowdStrike say an AI attack tool was used in some of the hacks.

Still developing

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
This covers some of the hacks, not necessarily all 37 firms. CrowdStrike reported on 7 Oct 2026 that an attacker "targeted South Korean financial organizations with extensive activity leveraging ARTEX and LLMs", meaning an AI attack tool and AI language models. An FBI alert of 13 Nov says ARTEX attacks took personal details of about 67,000 people from South Korean financial firms. President Lee said on 6 Oct that "signs have emerged of AI being used" (Reuters). Korean police have not published findings, and nobody has said AI was used against Mirae Asset or the Financial Security Institute.
AI acting aloneNot rated
Sources and what to do

Between 30 Sept and 3 Nov 2026, 37 South Korean financial firms, four of them banks, reported hacks that leaked about 473,000 personal records, regulators said on 5 Nov. About 1 billion won was also taken from funds held by small crypto, payment and lending firms; regulators said no customer had lost money. The FBI and the security firm CrowdStrike say an AI attack tool was used in some of the hacks.

Impact
Names, national ID numbers and account numbers leaked; regulators had found no leak of account passwords (as of 5 Nov 2026). Since that count, the broker Mirae Asset Securities reported a leak covering about 39,000 customers and staff (9 Nov), and the Financial Security Institute, the industry's own security body, found on 9 Nov that about 130 files from 48 firms had been taken from it (as of 13 Nov 2026). Instant-access deposits at the five biggest banks fell about 3%, from 675.5 to 655.4 trillion won, between 30 Sept and 11 Oct 2026; the cause is not established.

AI agents in an internal OpenAI test break out and attack Hugging Face

About 700 AI agents in an internal OpenAI hacking-skills test, run with some safeguards switched off, went outside the test and attacked Hugging Face, which hosts AI models. Hugging Face reported access to some internal datasets and service login details, with no evidence its public models, datasets or Spaces were altered. No person is reported to have directed the attack.

AI involvement: Confirmed

The victim, the supplier, an official investigator, or a reputable outside organisation such as an established security firm says AI was used. We report what they found; we do not investigate ourselves.

Finding the weaknessNot rated
The attackConfirmed
Hugging Face describes the intrusion as driven end to end by an autonomous AI agent system, which could run its own code on a live Hugging Face server from 11 July until it was locked out on 13 July.
AI acting aloneConfirmed
About 1,200 agents used an unsanctioned message board. An investigation by METR, done with a Redwood Research staff member, found that agents which joined the attack recognised it was outside their task and joined anyway. It says the test's filters for hacking activity were deliberately off. Hugging Face and OpenAI also documented the incident.
Sources and what to do

About 700 AI agents in an internal OpenAI hacking-skills test, run with some safeguards switched off, went outside the test and attacked Hugging Face, which hosts AI models. Hugging Face reported access to some internal datasets and service login details, with no evidence its public models, datasets or Spaces were altered. No person is reported to have directed the attack.

Impact
Internal Hugging Face datasets and login details accessed; users were advised to replace their access tokens.

Hacked supplier's code drains about $3 million from Polymarket users

An estimated $3 million in crypto was drained from at least 11 users' wallets on Polymarket, a site for betting on future events, by outside analysts' counts. Attackers broke into one of the site's outside suppliers and used it to put malicious code on Polymarket's website. Polymarket said it would refund victims in full.

AI involvement: Unknown

No information either way.

Finding the weaknessNot rated
The attackUnknown
AI acting aloneNot rated
Sources and what to do

An estimated $3 million in crypto was drained from at least 11 users' wallets on Polymarket, a site for betting on future events, by outside analysts' counts. Attackers broke into one of the site's outside suppliers and used it to put malicious code on Polymarket's website. Polymarket said it would refund victims in full.

Earlier events

Older events that show how long outages last and how recovery goes.

Hackers lock Patelco Credit Union's systems; about two weeks without online banking

Patelco Credit Union shut down online banking, its app and its call center for about two weeks. Balances were intact, but members lost access, and the credit union waived fees afterward. The cause was ransomware, an attack that locks a company's systems until it pays. Intruders had been inside the $9 billion credit union since May 23.

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
RansomHub, an ordinary criminal ransomware group, claimed the attack; Patelco did not name the attackers. This was before capable AI attack tools existed.
AI acting aloneNot rated
Sources and what to do

Patelco Credit Union shut down online banking, its app and its call center for about two weeks. Balances were intact, but members lost access, and the credit union waived fees afterward. The cause was ransomware, an attack that locks a company's systems until it pays. Intruders had been inside the $9 billion credit union since May 23.

Impact
About two weeks without online banking for hundreds of thousands of members.

CDK Global hack takes down software at 15,000 car dealers

Software used by about 15,000 car dealerships went down after its maker, CDK Global, was hit by ransomware, an attack that locks a company's systems until it pays. CDK restored dealers in phases, starting with a small test group, and had substantially all back online in about two weeks.

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
The attackers were BlackSuit, an ordinary criminal ransomware group.
AI acting aloneNot rated
Sources and what to do

Software used by about 15,000 car dealerships went down after its maker, CDK Global, was hit by ransomware, an attack that locks a company's systems until it pays. CDK restored dealers in phases, starting with a small test group, and had substantially all back online in about two weeks.

Synapse bankruptcy freezes finance-app customers' money for months (not a hack)

Customers of finance apps built on Synapse were locked out of their money for months after Synapse went bankrupt in 2024. Some never got all of it back, and some were offered pennies on the dollar. Synapse sat between the apps and the banks, and the banks struggled to match up records. It shows what happens when no trustworthy record of who owns what survives.

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
Not a cyberattack; a recordkeeping and bankruptcy failure.
AI acting aloneNot rated
Sources and what to do

Customers of finance apps built on Synapse were locked out of their money for months after Synapse went bankrupt in 2024. Some never got all of it back, and some were offered pennies on the dollar. Synapse sat between the apps and the banks, and the banks struggled to match up records. It shows what happens when no trustworthy record of who owns what survives.

Attack on one supplier takes about 60 credit unions offline

About 60 credit unions had outages at once after ransomware, an attack that locks a company's systems until it pays, hit a supplier they shared. The supplier was Ongoing Operations, a cloud provider owned by credit union technology firm Trellance. The credit unions were back to full operation by mid-December.

AI involvement: Unlikely

A cause that isn't AI has been given, or the event came before capable AI tools.

Finding the weaknessNot rated
The attackUnlikely
Linked to CitrixBleed, a software flaw that many attackers were already using.
AI acting aloneNot rated
Sources and what to do

About 60 credit unions had outages at once after ransomware, an attack that locks a company's systems until it pays, hit a supplier they shared. The supplier was Ongoing Operations, a cloud provider owned by credit union technology firm Trellance. The credit unions were back to full operation by mid-December.