Data breaches 2026: which recent breaches exposed passwords, and what to do
Most large data breaches of 2026 exposed names, emails and phone numbers, not passwords. Between August and November 2026, the ones that did expose passwords included Fanlore (MD5 or PBKDF2 hashes), Chainstack (password hashes) and the CastleRAT malware takedown (893,000 passwords). If you were in one, change that password everywhere you used it.
Key takeaways
- Of the 31 breaches Have I Been Pwned added between 1 August and 19 November 2026, 3 included passwords, and one of those was a 2013 breach loaded late. The biggest ones, such as Carhartt (12.9 million emails) and Exact Sciences (10.9 million), did not.
- Password-related incidents in this period include Fanlore (145,000 accounts, MD5 or PBKDF2 hashes), Chainstack (password hashes, detected 31 October 2026) and Capcom (196 accounts had passwords changed by an intruder).
- Europol's November 2026 takedown of the CastleRAT malware passed 893,000 unique passwords to Have I Been Pwned. Malware steals passwords from your device, so a strong password alone does not protect you.
- A breach without passwords still matters: leaked emails and phone numbers feed phishing. The 2026 FIDO Alliance survey found 33% of consumers had a confirmed compromise or breach notice in the past year.
- The fix is the same for every password breach: a unique, random password per site, stored in a password manager, plus a second factor or a passkey.
What counts as a data breach in 2026?
A data breach is any incident where personal data leaves the organisation that holds it without permission, whether through hacking, a misconfigured system, a malicious insider or a supplier. In 2026 the most common pattern in public disclosures was "pay or leak" extortion: criminals copy data, demand payment, and publish it if the company refuses.
Three related terms are worth separating, because they need different responses:
- Breach: data taken from a company's systems. You cannot prevent it, but you can limit the damage.
- Scrape: data collected from public or semi-public pages at scale. Have I Been Pwned treated the August 2026 Chess.com data as a likely scrape, because 99% of its email addresses had already appeared in earlier breaches.
- Infostealer log: data stolen from an infected computer, often including every password saved in its browser. This is the most dangerous for passwords, because the passwords are captured in plain text before any hashing happens.
Which data breaches in 2026 exposed passwords?
Between August and November 2026, the publicly confirmed incidents that exposed passwords or password hashes included Fanlore, Chainstack, the CastleRAT infostealer data and Capcom, where intruders changed some account passwords. The large retail and healthcare breaches of the same period exposed contact and health data, not passwords.
The table below lists incidents with an official disclosure or a Have I Been Pwned (HIBP) entry, sorted by what happened to passwords. Last checked 19 November 2026.
| Incident | Disclosed | Accounts | Passwords |
|---|---|---|---|
| CastleRAT malware (Europol) | Nov 2026 | 179,745 | 893,000 plain passwords from infected PCs |
| Fanlore (OTW) | Aug 2026 | 144,520 | MD5 or PBKDF2 hashes |
| Chainstack | 3 Nov 2026 | Not stated | Hashes, algorithm not stated |
| Capcom (CAPCOM ID) | Oct 2026 | More than 11 million | 196 changed by intruders; reset by Capcom |
| Carhartt | Aug 2026 | 12,933,413 | Not included |
| Exact Sciences | Aug 2026 | 10,869,543 | Not included (health data was) |
| Manchester Airports Group | Aug 2026 | 8,849,657 | Not included |
| McKesson | Aug 2026 | 6,404,340 | Not included (health data was) |
| Chess.com (likely scrape) | Aug 2026 | 4,653,212 | Not included |
| Oz Hair and Beauty | Aug 2026 | 1,988,331 | Not included |
| RingCentral | Jul to Aug 2026 | 1,596,490 | Not included |
| Brinks Home | Jul to Aug 2026 | 732,162 | Not included (partial card data was) |
| Medela | Sep 2026 | 423,947 | Not included |
| Double Counter | Oct 2026 | 274,922 | Not included |
Account counts are unique email addresses as loaded into HIBP, except Capcom, which comes from its disclosure as reported by Security Magazine, and Chainstack, which did not publish a number. We will update this table when companies publish final counts or when HIBP adds a breach from this period.
The password breaches in detail
Fanlore, August 2026
In August 2026, the Organization for Transformative Works found unauthorised access to the Fanlore wiki it runs. HIBP lists 144,520 unique email addresses with usernames and passwords "stored as either MD5 or PBKDF2 hashes". OTW submitted the data to HIBP itself, which is the responsible way to handle a breach.
The two hash types matter a great deal. PBKDF2 is a deliberately slow password hashing scheme. MD5 is a fast general-purpose hash that the OWASP Password Storage Cheat Sheet says was "designed to be fast", which makes it unsuitable for passwords. If your Fanlore password was stored as MD5 and it was short or common, assume it is known.
Chainstack, October to November 2026
Chainstack, a blockchain infrastructure provider, published an incident report on 3 November 2026, updated on 11 November. An attacker entered a development environment on 26 October 2026, reached parts of production on 29 October, and was detected and cut off on 31 October. The exposed database held organisation names, account holder names, member email addresses and password hashes. Chainstack did not say which hashing algorithm it used.
Chainstack's own advice: change your Chainstack password and any other password you reused, rotate node credentials, regenerate platform API keys, and watch for phishing. That list is a good model for any developer platform breach, because API keys and tokens are often more valuable to an attacker than the password.
Capcom, October 2026
Security Magazine reported on 22 October 2026 that Capcom disclosed unauthorised access affecting more than 11 million CAPCOM ID accounts. The exposed data was mostly account metadata: creation and last login times, number of logins and the last login IP address, with birthdate, nickname, gender and region for a small subset. For 196 accounts, attackers changed the login password. Capcom reset the affected passwords and warned users about phishing. The report did not say whether original passwords were exposed.
CastleRAT takedown, November 2026
In November 2026, Europol and partner agencies dismantled infrastructure behind several malware operations, including CastleRAT, a remote access trojan. HIBP added 179,745 email addresses associated with CastleRAT infections on 13 November 2026, along with 893,000 unique passwords seen more than 3 million times on infected computers. Those passwords are now searchable in Pwned Passwords.
This is the incident type that strong passwords cannot fix. Malware on your own computer captures what you type or what your browser stores. If your email appears in this set, clean the device first, then change passwords from a different, clean device.
Why most 2026 breaches did not include passwords
Most of the largest 2026 breaches came from customer relationship and marketing systems rather than login databases, so they held contact details, purchase histories and health records instead of passwords. HIBP describes Carhartt, Exact Sciences, McKesson, RingCentral, Brinks Home, Medela and Crunchbase as targets of ShinyHunters "pay or leak" extortion campaigns.
That is not good news for the people affected. Exact Sciences' data included personal health data, and Brinks Home's included partial card details (last four digits, type and expiry). Contact data is the raw material for phishing, and a convincing phishing email is one of the main ways passwords are stolen in the first place.
The Verizon 2026 Data Breach Investigations Report, covering incidents from 1 November 2024 to 31 October 2025, says software vulnerabilities have overtaken stolen passwords as the top way in, at 31% of breaches. It also found ransomware in 48% of breaches. Stolen credentials are no longer first, but they have not gone away.
What data did the 2026 breaches expose most often?
Email addresses were exposed in every one of the 31 breaches Have I Been Pwned added between 1 August and 19 November 2026, names in 25 and phone numbers in 22. Passwords appeared in only 3. We counted these ourselves from the data classes HIBP publishes for each breach.
| Data exposed | Breaches | Share of 31 |
|---|---|---|
| Email addresses | 31 | 100% |
| Names | 25 | 81% |
| Phone numbers | 22 | 71% |
| Physical addresses | 18 | 58% |
| Dates of birth | 10 | 32% |
| Employers | 10 | 32% |
| Job titles | 9 | 29% |
| Usernames | 5 | 16% |
| Purchases | 4 | 13% |
| Passwords | 3 | 10% |
| Government issued IDs | 3 | 10% |
| Partial card data | 3 | 10% |
This mix explains the advice that follows. A leaked email, phone number, address and employer is a ready-made kit for a believable phishing message: "Hi Sam, about your order delivered to Elm Street, please confirm your login." The phishing page then asks for the password that was not in the breach. Fourteen of the 31 breaches, covering 37.2 million accounts between them, were attributed by HIBP to the same ShinyHunters "pay or leak" campaigns.
"Added in 2026" is not the same as "happened in 2026"
Eight of the 31 breaches HIBP loaded between August and November 2026 actually happened in earlier years, from 2013 to 2025. Old breaches keep surfacing because stolen data is resold, merged into larger collections and eventually leaked publicly, sometimes years later.
| Breach | Happened | Loaded | Accounts |
|---|---|---|---|
| LimeLeads | 2019 | Sep 2026 | 17,838,396 |
| YouHack | 2013 | Nov 2026 | 7,160,849 |
| Angel One | 2023 | Oct 2026 | 6,765,054 |
| SberSpasibo | 2023 | Nov 2026 | 3,288,711 |
| Burger King Russia | 2024 | Sep 2026 | 3,155,792 |
| Harvard | 2025 | Nov 2026 | 850,537 |
Two practical points follow. First, a password you used years ago and forgot about can still appear in a "new" breach, so retire old accounts you no longer use. Second, the largest single entry added in this period, LimeLeads with 17.8 million addresses, came from an exposed database in 2019. Across the 31 breaches, 93.5 million accounts were added, but only 54.4 million came from breaches that happened in 2026.
Data breach statistics for 2026
The 2026 figures point the same way: breaches are more expensive, more of them are extortion, and password failures remain common for ordinary people. Here are the numbers we could trace to a primary source.
| Figure | Value | Source and year |
|---|---|---|
| Average global cost of a breach | US$4.99m | IBM and Ponemon, 2026, up 12% |
| Breaches starting with a software vulnerability | 31% | Verizon DBIR 2026 |
| Breaches involving ransomware | 48% | Verizon DBIR 2026 |
| Consumers with a compromise or breach notice in the past year | 33% | FIDO Alliance survey, April 2026 |
| Same figure, United States | 41% | FIDO Alliance survey, April 2026 |
| Consumers likely to abandon a purchase over a forgotten password | 47% | FIDO Alliance survey, April 2026 |
| Password attacks observed per second | 7,000 | Microsoft, 2024 data, published May 2025 |
| Breaches in HIBP | 1,052 | HIBP API, 19 November 2026 |
| Breached accounts in HIBP | 17,858,376,246 | HIBP API, 19 November 2026 |
| HIBP breaches that include passwords | 676 | HIBP API, 19 November 2026 |
Two of these deserve context. The FIDO Alliance figures come from a Sapio Research survey of 11,000 consumers in ten countries in April 2026, with a margin of error of plus or minus 0.9 percentage points. Microsoft's figure of 7,000 password attacks per second refers to what it observed during 2024, more than double the 2023 rate.
The HIBP numbers are our own count from its public breach list. Since HIBP started, 676 of its 1,052 breaches (64%) have included passwords in some form. Between 1 August and 19 November 2026, only 3 of 31 added breaches did, about 10%, and one of the three (YouHack) was a 2013 breach of a Russian-language hacking forum that HIBP loaded in November 2026, with mostly unsalted MD5 hashes. Our reading: password databases are better protected than they were, and attackers are going after other data or straight to the device.
Which sectors were hit, and what that means for you
Healthcare and business software were the most frequent targets among breaches that happened in 2026 and were loaded into HIBP between August and November. Each sector leaks a different kind of data, so the follow-up is different too.
| Sector | 2026 examples | Typical data | Your main risk |
|---|---|---|---|
| Healthcare | Exact Sciences, McKesson, Sharecare, Medela, Alcon, Neogen | Health records, dates of birth, IDs | Targeted scams and identity fraud |
| Business software | RingCentral, Crunchbase, Questel, CyrusOne | Work contacts, support tickets | Business email compromise |
| Retail | Carhartt, Oz Hair and Beauty, Brinks Home | Addresses, purchases, partial cards | Fake delivery and refund messages |
| Travel | Manchester Airports Group, Gites de France | Bookings, vehicle plates | Fake booking and parking emails |
| Developer and online services | Chainstack, Double Counter, Fanlore | Password hashes, usernames, keys | Account takeover, stolen API keys |
For a health data breach, a password change does little, because the harm is the data itself. Watch for messages that quote your medical details to gain trust, and be wary of anyone asking you to "confirm" insurance or identity information. For a business software breach, warn colleagues: a leaked support ticket can be quoted back to you in a convincing fake invoice.
Is a hashed password safe after a breach?
A hashed password is safe after a breach only if the hash is a slow, salted password hash and the password itself is long and random. A short or common password can be recovered from almost any hash by guessing, and a fast hash such as MD5 makes that guessing cheap.
Hashing turns a password into a fixed string that cannot be reversed directly. But an attacker with the hashes can try billions of guesses offline, hash each one, and look for matches. How fast that goes depends on the algorithm. The OWASP Password Storage Cheat Sheet recommends these minimums, each designed to make every guess expensive:
| Algorithm | OWASP minimum | Verdict |
|---|---|---|
| Argon2id | 19 MiB memory, 2 iterations, 1 thread | Preferred |
| scrypt | Cost 2^17, block size 8, parallelism 1 | Good |
| bcrypt | Work factor 10 or more, 72 byte limit | Legacy systems |
| PBKDF2-HMAC-SHA256 | 600,000 iterations | Acceptable |
| MD5, SHA-1, plain SHA-256 | Not for passwords | Unsafe |
Here is a worked comparison using the guessing rate this site uses for its own crack-time tables: 10 billion guesses a second against a fast, unsalted hash. A random 8 character password using about 87 possible characters has roughly 51.5 bits of entropy. On average it falls after half the possibilities are tried, about 2^50.5 guesses, which takes about 44 hours at that rate. A random 16 character password from the same set has about 103 bits, and the same attacker would need far longer than the age of the universe. Against a slow hash such as Argon2id or bcrypt, each guess costs thousands of times more, which helps the short password a little and makes the long one irrelevant to guess.
The practical rule: you cannot choose how a company stores your password, so choose a password that survives the worst case. Sixteen random characters from the 16 character generator survive an MD5 leak. A pet's name and a year do not survive any of them.
What should I do if my password was in a data breach?
If your password was in a data breach, change it on that site, then change it on every other site where you used the same or a similar password, starting with your email. Then turn on a second factor or a passkey where the site offers one.
What you do first depends on what leaked. Use this table to decide.
| What leaked | Next hour | Next day | Next week |
|---|---|---|---|
| Plain-text password or infostealer log | Clean the device; change email password from a clean device | Change every saved password; sign out all sessions | Move to a password manager and passkeys |
| Password hashed with MD5 or SHA-1 | Change it on that site and anywhere reused | Turn on 2FA on email, bank, cloud | Replace any password under 15 characters |
| Password hashed with bcrypt, scrypt, Argon2 or PBKDF2 | Change it on that site | Change it where reused | Check for other reuse |
| Hashes, algorithm not stated | Treat as MD5: change it everywhere used | Turn on 2FA | Replace reused passwords |
| Email, phone, address only | Nothing urgent | Expect phishing that uses the leaked details | Check bank and card statements if payment data was involved |
| API keys or tokens | Revoke and regenerate | Review access logs | Rotate any related secrets |
A step-by-step recovery checklist
Work through these steps in order. The order matters, because your email account can reset every other account, and a compromised device can capture every new password you type.
- Check the device. If your email appears in an infostealer or malware dataset such as CastleRAT, run a full scan or reset the device before changing anything.
- Secure your email first. Make a new, unique password of at least 16 random characters with the password generator. Turn on two-factor authentication or add a passkey.
- Sign out of all sessions on your email account, so anyone already signed in is removed.
- Change the breached site's password. If the site has closed accounts or forced a reset, follow its process rather than links in emails.
- Find every reuse. List the sites where you used the same password or a variation of it. Variations count: "Summer2025!" and "Summer2026!" are the same password to a cracking tool.
- Replace each one with a different random password, saved in a password manager. Our Bitwarden vs 1Password comparison covers two popular choices.
- Add passkeys on accounts that support them. A passkey cannot be reused or phished the way a password can. See passkey vs password for how they compare.
- Check the new passwords against known breaches with the strength checker, which sends only part of a hash, never the password.
- Watch for phishing for several months. Breach data is used for targeted emails that mention real details, such as your order history or the airport car park you booked.
- Revoke and rotate any API keys, tokens or app passwords for the breached service.
Credential stuffing: why one leak opens other accounts
Credential stuffing is an attack where criminals take email and password pairs from one breach and try them automatically on many other sites. It works because people reuse passwords, so a leak from a small forum can open a bank or email account.
Credential stuffing does not need any clever cracking. If the leaked password was stored in plain text, or recovered from a weak hash, the attacker simply replays it. Microsoft's figure of 7,000 password attacks a second in 2024 gives a sense of the scale of automated guessing against just one company's accounts.
The defence is entirely on your side: one password, one site. If every password is unique, a breach anywhere costs you one account, the one that was breached. NIST's SP 800-63B-4, finalised on 31 July 2025, also requires sites to check new passwords against lists of compromised values, so a password that has already leaked should be refused when you try to set it. Our guide to the NIST password guidelines covers the rest of those rules.
How do I check if I have been in a data breach?
To check whether your email address has been in a data breach, search it on haveibeenpwned.com. To check whether a specific password has leaked, use a tool that sends only a short prefix of the password's hash, such as the breach check in our strength checker.
Those are two different questions. An email search tells you which breaches included your address. A password check tells you whether that exact password appears anywhere in breach data, even from a breach that never included your email. A password found there should never be used again, even if it looks strong.
The password check here uses the Have I Been Pwned range API. Your browser hashes the password with SHA-1, sends only the first 5 of the 40 characters, and compares the rest locally against the list that comes back. We explain the method step by step in how to check if your password was leaked.
Rule of thumb: if a breach notice arrives, do not click its links. Go to the company's site by typing the address, or use its app. Breach notices are copied by phishing campaigns within days.
Who just had a data breach? How to keep track
The most reliable way to track new data breaches is Have I Been Pwned's notification service, which emails you when your address appears in a newly loaded breach, plus official notices from companies you use. News roundups are useful but often repeat early, unconfirmed figures.
We built the table on this page from primary sources only: HIBP's public breach list, which describes each breach and what data it held, and the companies' own incident reports where they exist. Several large 2026 figures that circulate online, including the AI-generated summaries that appear at the top of search results, could not be traced to a primary source, so we left them out.
Signs that a breach report is reliable:
- It links to the company's own notice, a regulator filing or HIBP.
- It separates unique email addresses from total rows of data.
- It says whether passwords were included and how they were stored.
- It distinguishes a confirmed breach from a claim by a criminal group.
Lessons for developers and IT teams
The 2026 incidents give site owners a short, concrete list: store passwords with a slow hash, screen new passwords against breach lists, offer a phishing-resistant sign-in, and treat development systems and analytics tools as production.
- Development is production. Chainstack's attacker started in a development environment on 26 October 2026 and reached production three days later. Secrets and network paths shared between the two turn a minor intrusion into a customer breach.
- Patch internal tools. Double Counter attributed its October 2026 breach to a vulnerability in the Metabase analytics tool. Dashboards and admin panels often hold full copies of customer data with weaker protection than the main app.
- Hash with Argon2id, scrypt, bcrypt or PBKDF2 at OWASP strength. Fanlore's mix of MD5 and PBKDF2 hashes shows how long legacy hashes survive. OWASP recommends rehashing to a modern algorithm the next time each user signs in.
- Screen passwords at creation. NIST SP 800-63B-4 requires verifiers to compare new passwords against a blocklist of commonly used, expected or compromised values. The Pwned Passwords range API does this without receiving the password.
- Offer a phishing-resistant option. NIST SP 800-63B-4 says verifiers shall offer at least one phishing-resistant authentication option at AAL2. Passkeys meet that.
- Plan the notice before you need it. Chainstack's report gave a timeline, the exact data types and specific actions for customers. That clarity is what lets people protect themselves quickly.
A 15 minute plan to prepare for the next breach
You cannot stop the next breach, but you can make sure it costs you one password change instead of a weekend of account recovery. This plan takes about 15 minutes for the accounts that matter most.
- List your top five accounts: main email, phone account (Apple, Google or Microsoft), bank, password manager and the place you store files or photos.
- Give each a unique random password of at least 16 characters, or a six word passphrase for the ones you type by hand.
- Turn on a second factor on all five. An authenticator app or a passkey is stronger than a text message code.
- Add a passkey where offered. Many large email and platform accounts now support them.
- Save recovery codes somewhere offline, such as printed in a drawer.
- Subscribe your email to Have I Been Pwned's notification service so you hear about new breaches early.
- Delete accounts you no longer use. An old shopping or forum account is a breach waiting to happen, and old breaches keep resurfacing.
After that, the rest of your accounts can be moved over gradually, each time you log in to one. A password manager makes this practical, because it remembers every unique password for you.
What this means for the passwords you make today
The 2026 breaches lead to three rules for any new password: make it random, make it long, and make it unique. A generator does the first two; a password manager makes the third practical.
- Random: people are predictable. A password you invent follows patterns that cracking tools try first. Let a generator choose, such as the password generator on this site, which runs in your browser and never sends what it makes.
- Long: 16 characters for anything stored in a password manager. NIST requires at least 15 for a password used alone. For one you must remember, use a passphrase of five or six random words.
- Unique: never reuse. This one rule turns every future breach from a crisis into a single password change.
And where you can, replace the password with a passkey. A passkey leaves nothing reusable on the company's server, so the next breach of that site does not expose a secret that works anywhere else.
Questions
What are the biggest data breaches of 2026 so far?
Among breaches loaded into Have I Been Pwned between August and November 2026, the largest were Carhartt (12.9 million email addresses), Exact Sciences (10.9 million), Manchester Airports Group (8.8 million) and McKesson (6.4 million). None of those four included passwords.
Who just had a data breach recently?
Recent incidents disclosed in October and November 2026 include Capcom (CAPCOM ID accounts), Chainstack (password hashes, detected 31 October 2026), Double Counter and the CastleRAT malware data released through Europol. Check Have I Been Pwned for the newest additions.
How do I check if I have been in a data breach?
Search your email address on Have I Been Pwned to see which breaches included it. To check a specific password, use a k-anonymity check like the one in our strength checker, which never sends the password itself.
Should I change my password after a data breach?
Yes, if the breach included passwords or password hashes, or if you are unsure. Change it on the breached site and on every site where you used the same or a similar password, starting with your email.
Is my password safe if it was hashed?
Only if the site used a slow password hash such as Argon2id, scrypt, bcrypt or PBKDF2 and your password was long and random. Short or common passwords can be recovered from most hashes, and fast hashes such as MD5 make it easy.
What is credential stuffing?
It is an automated attack that takes email and password pairs from one breach and tries them on other sites. It only works when people reuse passwords, which is why a unique password for every site is the main defence.
Can a strong password protect me from infostealer malware?
No. Infostealer malware captures passwords on your own device, however strong they are. Keep devices updated, avoid pirated software and unknown downloads, and use passkeys, which malware cannot replay on another site.
Do I need to do anything if only my email address leaked?
Nothing urgent, but expect more phishing. Criminals use leaked details such as order history or bookings to make fake emails convincing. Do not click links in breach notices; go to the company site directly.
Sources
- Have I Been Pwned breach list and API (breaches added August to November 2026), Have I Been Pwned. Checked 19 November 2026.
- Incident report: what happened and what customers need to do, Chainstack. Checked 19 November 2026.
- 11M impacted by Capcom breach, some facing unauthorized changes, Security Magazine. Checked 19 November 2026.
- Chainstack reports production systems breach exposing customer data and password hashes, The Hacker News. Checked 19 November 2026.
- 2026 Data Breach Investigations Report, Verizon Business. Checked 19 November 2026.
- Cost of a Data Breach Report 2026, IBM. Checked 19 November 2026.
- The State of Passkeys 2026: Global Consumer and Workforce Report, FIDO Alliance. Checked 19 November 2026.
- Pushing passkeys forward: Microsoft’s latest updates for simpler, safer sign-ins (1 May 2025), Microsoft Security Blog. Checked 19 November 2026.
- Password Storage Cheat Sheet, OWASP. Checked 19 November 2026.
- SP 800-63B-4, section 3.1.1.2: Password Verifiers, National Institute of Standards and Technology. Checked 19 November 2026.