State breach notices to attorneys general, 2012 to 2026: more notices, ransomware, and longer gaps
We counted every notice on the published Washington, California and Oregon lists to see how many breaches reach the states, how large they are and what causes them.
Three states, three lists, two thresholds
This report uses three states whose attorney general lists we could download and count in full on November 16, 2026, each as one export or one table: Washington, California and Oregon. We also tried Maine's published list, but its pages returned errors to our requests that day, so Maine is left out rather than counted in part.
- Washington: a business that must notify more than 500 Washington residents of a single breach must notify the attorney general no more than 30 days after the breach was discovered, giving the number of Washington consumers affected, the types of information, the time frame of exposure and a sample notice [4]. That 30-day rule came from the 2019 amendment, which took effect on March 1, 2020 [4]. The Attorney General's Office publishes statistics derived from these notices on the state's open data portal, the same data behind its annual data breach report [1].
- California: a business or agency that sends a breach notice to more than 500 California residents must give a sample copy to the attorney general, who lists those sample notices with the organization's name, the date or dates of breach if known, and the date reported [2].
- Oregon: a covered entity must notify the attorney general when the number of consumers it must notify exceeds 250 [5], and must notify consumers no later than 45 days after discovering or being told of the breach [5]. The Department of Justice lists the notices it receives with the dates of breach, discovery and notice and a number affected [3].
Washington and California share a threshold of more than 500 of their own residents, while Oregon's is more than 250 [4] [2] [5], so Oregon's list takes in smaller incidents and its counts are not directly comparable with the other two. In all three lists the name shown is the organization that sent the notice, which may not be the one that was breached: California and Oregon give the example of a bank notifying customers of a card breach at a merchant [2] [3], and Washington's data dictionary says the notifier is nearly always the owner of the data even when its contractor was breached [1].
How we counted
All three sources were downloaded on November 16, 2026.
- Washington: the dataset "Data Breach Notifications Affecting Washington Residents" (sb4j-ca4h) on data.wa.gov, exported as CSV with 1,698 rows, one per notice, submitted from August 11, 2015 to November 9, 2026 [1]. We counted each row in the calendar year of its DateSubmitted field. The dataset's own Year field runs from July 24 to July 23, because of the effective date of the law requiring notice to the attorney general, so our yearly counts differ from the ones the Attorney General's Office reports [1]. "Residents affected" is the WashingtoniansAffected field, defined as the estimated or known number of Washington residents whose information was affected [1]; 32 rows leave it blank, and medians and sums use only the rows with a number [1]. Cause is the DataBreachCause field (Cyberattack, Unauthorized Access, Theft or Mistake) and attack type the CyberattackType field, which the data dictionary describes as a sub-category of cyberattacks [1]; eight rows with another cause also carry an attack type [1], and we counted attack types on every row that has one, whatever its cause.
- California: the "Download Full Data Breach List (CSV)" file from the attorney general's list, 5,440 rows reported from January 20, 2012 to November 13, 2026 [2]. We counted each row in the calendar year of its Reported Date. The list carries no count of people affected [2], so California figures here are counts of notices only; 496 rows give no breach date [2].
- Oregon: the Department of Justice's search page, which shows its full table when opened without a search; we read the table's 1,682 rows, reported from October 30, 2015 to October 29, 2026 [3]. We counted each row in the calendar year of its Reported Date. The page does not say whether its Number Affected column counts Oregon consumers or everyone notified, and says it shows the figures from an entity's first report, without later updates [3]; 199 rows leave it blank [3]. We therefore use Oregon for counts of notices, and quote its Number Affected only as listed.
- Duplicates: we removed none. In California, 32 sets of rows share an organization name and reported date (41 extra rows), and 12 extra rows repeat another row exactly [2]; in Washington, six pairs share a name (ignoring capitalization) and submission date [1]; in Oregon, 21 extra rows share a name and reported date [3]. Separate notices from one organization can be legitimate (different affiliates or incidents), so we kept every row as the state lists it.
- Timing: in Washington, DaysElapsedBeforeNotification is the number of days from DateAware, when the notifier became aware of the breach, to DateSubmitted [1]; we confirmed the field equals that difference on all 1,675 rows that give it [1]. In California, we measured from the latest breach date a row lists to its Reported Date, for the rows that give a breach date [2].
Limits. These are lists of notices, not of incidents: one incident at a shared vendor can produce many notices, and a resident caught in several breaches is counted once per notice. A state list shows only what was sent to that state, and recent weeks fill in late. Figures describe each source as downloaded on November 16, 2026; percentages and medians are rounded.
Notices by year
| Year | California notices | Washington notices | Oregon notices | Washington residents affected (sum) | Washington median per notice | Washington notices of one hundred thousand residents or more | Cyberattack | Ransomware | Unauthorized access | Theft or mistake |
|---|---|---|---|---|---|---|---|---|---|---|
| 2015 (WA from Aug 11; OR from Oct 30) | 211 [2] | 14 [1] | 2 [3] | 380,447 [1] | 2,721 [1] | 1 [1] | 50% [1] | 0% [1] | 29% [1] | 21% [1] |
| 2016 | 244 [2] | 49 [1] | 70 [3] | 2,092,413 [1] | 1,409 [1] | 2 [1] | 51% [1] | 2% [1] | 39% [1] | 10% [1] |
| 2017 | 317 [2] | 83 [1] | 103 [3] | 4,249,445 [1] | 1,236 [1] | 2 [1] | 65% [1] | 0% [1] | 25% [1] | 10% [1] |
| 2018 | 262 [2] | 63 [1] | 108 [3] | 299,284 [1] | 1,607 [1] | 0 [1] | 70% [1] | 0% [1] | 19% [1] | 11% [1] |
| 2019 | 252 [2] | 60 [1] | 109 [3] | 562,218 [1] | 1,684 [1] | 1 [1] | 75% [1] | 7% [1] | 17% [1] | 8% [1] |
| 2020 | 390 [2] | 204 [1] | 115 [3] | 4,789,657 [1] | 1,944 [1] | 12 [1] | 80% [1] | 57% [1] | 15% [1] | 5% [1] |
| 2021 | 479 [2] | 186 [1] | 179 [3] | 5,657,355 [1] | 1,643 [1] | 7 [1] | 81% [1] | 34% [1] | 13% [1] | 6% [1] |
| 2022 | 455 [2] | 171 [1] | 165 [3] | 4,260,467 [1] | 1,704 [1] | 8 [1] | 64% [1] | 33% [1] | 30% [1] | 6% [1] |
| 2023 | 642 [2] | 252 [1] | 236 [3] | 11,494,295 [1] | 2,482 [1] | 15 [1] | 69% [1] | 33% [1] | 27% [1] | 4% [1] |
| 2024 | 608 [2] | 228 [1] | 185 [3] | 7,217,047 [1] | 1,940 [1] | 8 [1] | 76% [1] | 48% [1] | 19% [1] | 5% [1] |
| 2025 | 565 [2] | 209 [1] | 248 [3] | 6,108,180 [1] | 2,088 [1] | 12 [1] | 75% [1] | 45% [1] | 23% [1] | 2% [1] |
| 2026 (to Nov 13 CA, Nov 9 WA, Oct 29 OR) | 527 [2] | 179 [1] | 162 [3] | 2,564,692 [1] | 2,226 [1] | 5 [1] | 66% [1] | 38% [1] | 31% [1] | 3% [1] |
The last four columns are shares of Washington notices in the year. Oregon counts notices to more than 250 residents and the other two more than 500 [5] [4] [2], so the comparisons that follow are of each state's change over time, not of one state's level against another's. California's list grew from 1,465 notices reported in 2016 to 2020 to 2,749 in 2021 to 2025, an increase of 88%, with its highest year in 2023 at 642 [2]. Oregon's doubled over the same two five-year spans, from 505 to 1,013 [3], and Washington's rose from 459 to 1,046 [1], though Washington's jump in 2020 coincides with the amendment that set a 30-day deadline from March 1, 2020 [4].
The most recent year runs differently in each state. California lists 527 notices reported from January 1 to November 13, 2026, against 484 in the same window of 2025 [2]; Washington 179 to November 9, against 178 [1]; Oregon 162 to October 29, against 216 [3].
Size: most notices are small, a few are very large
From 2021 to 2025, Washington's 1,046 notices reported 34,737,344 residents affected in total, counted once per notice [1]. The median notice in each of those years affected between 1,643 and 2,482 Washington residents [1]. Of the notices with a count, 28% affected fewer than 1,000 Washington residents and 24% affected 10,000 or more [1]. A handful of notices carry most of the total in some years: the five largest held 70% of residents affected in 2024 and 43% in 2025 [1]. Over the whole dataset, the ten largest of 1,698 notices hold 38% of the 49,675,500 residents affected [1].
Oregon's list, read with its caveat that the Number Affected column is undefined and not updated, shows 36 notices of 1,000,000 or more in 2023, 22 in 2024, 20 in 2025 and 17 in 2026 to October 29 [3], against four or five a year from 2016 to 2018 [3].
Causes: cyberattacks lead, and ransomware is the largest named type
Cyberattacks were the cause of 763 of Washington's 1,046 notices from 2021 to 2025 (73%), and account for 75% of residents affected in those years [1]. Ransomware was the attack type on 407 of the 1,046 notices (39%), ahead of malware (113) and phishing (73), while 133 carry the catch-all type Other [1]. Unauthorized access, which the dataset defines as a person deliberately reaching data through means such as an unsecured network or documents left out, was the cause of 233 [1], and theft or mistake of 50 [1].
The ransomware share peaked at 57% of notices in 2020 [1]. Of the 114 ransomware notices submitted in the second half of that year, 58 came from nonprofits and 35 from education organizations [1], a pattern one incident at a shared service provider could produce, though the dataset does not record whether a notice stems from a vendor incident [1]. The "Other" attack type rose from 16 notices in 2022 to 59 in 2023 [1].
Time to notify the attorney general
The gap between an organization learning of a breach and its notice reaching the state has widened. In Washington, the median number of days from DateAware to DateSubmitted was 36 in 2020, 83 in 2022, 134 in 2025 and 99 in 2026 to date, and the share of notices submitted more than 30 days after DateAware rose from 59% in 2020 to 87% in 2025 [1]. In California, the median from the latest listed breach date to the reported date was 44 days in 2012, 110 in 2020 and 174 in 2025 [2]; that measure includes the time to detect a breach as well as to report it.
These figures are not findings of late notice. Washington's 30-day clock runs from when the breach was discovered [4], and the dataset's DateAware is the date the notifier reports becoming aware [1], which can differ, for example when a data owner first hears of an incident from a vendor before knowing its residents were affected. The law also lets notice wait for a law enforcement request, and lets consumer notice wait for measures needed to determine the scope of the breach [4].
Moraph's view
Moraph's view: the three lists, each with its own threshold, agree on direction. Notice counts roughly doubled between the second half of the 2010s and the first half of the 2020s, the median notice stayed small, and a few very large notices drive the residents-affected totals, so a state's annual sum says more about one or two incidents than about the year. Ransomware has been the largest named attack type in Washington's data in every year since 2020, and we expect that to hold. The widening gap between awareness and notice is the change we would watch most closely: it points to more incidents reaching an organization through a vendor, and to longer investigations before an organization can say whose data was taken. We do not read the 2026 dip in Oregon or the flat Washington count as a turn; both lists fill in late.
What this means for you
- CISOs and security teams: Plan for ransomware as the most likely named cause of a reportable breach: it was the attack type on 39% of Washington notices from 2021 to 2025 [1]. Recovery and data-exfiltration controls matter as much as prevention.
- Privacy officers and general counsel: Keep a register of each state's attorney general threshold and clock; the three states here use more than 500, more than 500 and more than 250 residents [4] [2] [5], and Washington's notice is due no more than 30 days after discovery [4]. Record when and how discovery was established, since lists such as Washington's publish the gap from awareness to notice [1].
- Third-party risk teams: Ask vendors how fast they will tell you of an incident and whose data it touched. The lists name the notifier, often the data owner rather than the breached vendor [1] [2], so a vendor's delay becomes your delay on a public record.
- Boards and audit committees: Benchmark on counts and causes rather than on residents affected, which a few large notices dominate: ten of Washington's 1,698 notices hold 38% of all residents affected in the dataset [1].
Moraph predictions
Every prediction is Moraph's view, not a fact. Each has a date, a confidence level (70% means we expect to be wrong about three times in ten) and a test anyone can apply. We score them on this page when the date comes and never remove one. How we make and score predictions.
P1 By Confidence 80% Open
The California attorney general's data breach list shows at least 500 notices with a Reported Date in calendar year 2027.
How we will check: On or after March 31, 2028, download the full list from oag.ca.gov/privacy/databreach/list with its CSV link and count rows whose Reported Date falls in 2027. Right if 500 or more.
Basis: The list shows 455 notices reported in 2022, 642 in 2023, 608 in 2024, 565 in 2025 and 527 in 2026 through November 13 [2].
P2 By Confidence 80% Open
Cyberattack is the DataBreachCause on at least 60% of notices with a 2027 DateSubmitted in Washington's data breach notifications dataset on data.wa.gov.
How we will check: On or after March 31, 2028, export dataset sb4j-ca4h from data.wa.gov as CSV, keep rows whose DateSubmitted falls in calendar 2027, and divide rows with DataBreachCause Cyberattack by all such rows. Right if 60% or more.
Basis: The share was 80% in 2020, 81% in 2021, 64% in 2022, 69% in 2023, 76% in 2024, 75% in 2025 and 66% in 2026 through November 9 [1].
P3 By Confidence 75% Open
Ransomware is the CyberattackType on at least 30% of all notices with a 2027 DateSubmitted in Washington's data breach notifications dataset.
How we will check: On or after March 31, 2028, export dataset sb4j-ca4h as CSV, keep rows with a DateSubmitted in calendar 2027, and divide rows whose CyberattackType is Ransomware by all such rows (all causes). Right if 30% or more.
Basis: Ransomware's share of all notices was 57% in 2020, 34% in 2021, 33% in 2022, 33% in 2023, 48% in 2024, 45% in 2025 and 38% in 2026 through November 9 [1].
P4 By Confidence 75% Open
For notices with a 2027 DateSubmitted in Washington's dataset, the median DaysElapsedBeforeNotification is 60 days or more.
How we will check: On or after March 31, 2028, export dataset sb4j-ca4h as CSV, keep rows with a DateSubmitted in calendar 2027 and a value in DaysElapsedBeforeNotification, and take the median of that field. Right if 60 or more.
Basis: The median was 70.5 days in 2021, 83 in 2022, 67 in 2023, 109.5 in 2024, 134 in 2025 and 99 in 2026 through November 9 [1].
P5 By Confidence 75% Open
For rows of the California attorney general's breach list with a Reported Date in 2027 and at least one breach date, the median number of days from the latest listed breach date to the Reported Date is 100 or more.
How we will check: On or after March 31, 2028, download the CSV, keep rows with a 2027 Reported Date and at least one date in the Date(s) of Breach column, compute Reported Date minus the latest breach date in days for each, and take the median. Right if 100 or more.
Basis: Measured the same way, the median was 110 days in 2020, 109 in 2021, 132 in 2022, 114 in 2023, 160 in 2024, 174 in 2025 and 136 in 2026 through November 13 [2].
Facts on this page link to their sources. Text marked "Moraph's view", and every prediction, is our own analysis, not a fact or advice. Methodology.
Spotted an error? Tell us and we will check it and correct the piece where needed.