Corrections
Every material correction to a benefit figure or rule on this site, with the date it was made and the primary source that settled it.
A leave plan gets printed out and taken to an HR department. If this site published a wrong number, the people who relied on it are entitled to know that it happened, when, and whether they were affected — so the record is public rather than buried in a commit history.
Routine annual rate updates are not listed here; keeping current with a state's published figures is maintenance, not a correction. What follows is the shorter and more useful list: the times a figure or rule on this site was wrong, and what the issuing agency actually said. Each entry notes whether the error overstated or understated the benefit, because those are not equivalent to someone planning around a due date.
Found something that still looks wrong? Send it with a link to the primary source. That is the fastest route to a fix, and it will appear here.
The record
-
Washington's leave was summarised as ten weeks longer than the state will pay
What happened. The state list on this site's front page gave Washington as “26 wks”. Washington caps combined medical and family leave at 16 weeks in a claim year, or 18 when a pregnancy-related serious health condition causes incapacity. 26 weeks is Massachusetts's combined cap, and Washington's row sat directly beneath Massachusetts's in the same list. The figure was wrong from 13 May 2026 until 7 August 2026 — 86 days. Every other page that stated Washington's cap said 16 throughout, so the front page spent three months contradicting four of this site's own pages.
Who it affected. Overstated, by ten weeks. A Washington reader who took the front-page summary at face value could have sketched a leave plan around ten weeks of benefits the state does not provide. The calculator itself was never wrong: run for Washington it returns 16 weeks, and it did so the entire time. Only the at-a-glance summary was wrong, so anyone who actually ran their numbers saw the correct figure.
Primary source. RCW 50A.15.020: an eligible employee is entitled to “a combined total of 16 times the typical workweek hours”, extended to “18 times the typical workweek hours if the employee experiences a serious health condition with a pregnancy that results in incapacity”.
-
Washington was described as protecting jobs at employers of any size
What happened. The methodology page listed Washington alongside Massachusetts, Connecticut, Minnesota, Colorado, Oregon and Maine as states with “built-in state-level job protection at any employer size”. Washington's job protection reaches employers with 25 or more employees and requires 180 days of service. It has never applied at any employer size — before January 2026 the threshold was 50 employees, and HB 1213 lowered it to 25. The sentence was live from 12 May 2026 until 7 August 2026 — 87 days — while the front page of the same site correctly said “job protection expanded to 25+ employers Jan 2026”.
Who it affected. Overstated, and in the way that matters most. A woman at a Washington employer with fewer than 25 staff, or with under six months of service, was told she had a statutory right to return to her job. She does not. Her leave may be paid and still leave her employer free to replace her. The calculator's own output stated the 25-employee and six-month rules correctly throughout; it was the methodology page's summary of which states protect universally that was wrong.
Primary source. Washington State Employment Security Department, on the rules effective 1 January 2026: job protection applies at employers with “25 or more employees” and requires “180 calendar days (6 months)” of employment, the minimum-hours requirement having been removed. Codified at RCW 50A.35.
-
New Jersey's expanded job-protection thresholds were live law for ten days before this site said so
What happened. The New Jersey Family Leave Act's coverage thresholds dropped to 15 employees, 3 months of tenure and 250 hours worked when A3451 took effect on 17 July 2026. Until 27 July this site still presented the previous 30 employees / 12 months / 1,000 hours as current, describing the expansion as something still to come. The change had been researched and verified in May and entered as a future-dated note — and then the date arrived with nothing watching for it.
Who it affected. Understated. A worker at an employer with 15 to 29 employees, or with between 3 and 12 months of tenure, was told she had no state job protection during leave when in fact she had had it since 17 July. No payment figure was affected — only the job-protection wording.
Primary source. New Jersey Division on Civil Rights, the agency that enforces the Act: a covered employee is “a person who works for a company or organization with 15 or more employees worldwide, or a state or local government agency” who “has been employed by the employer for at least 3 months, and has worked at least 250 hours in the past 12 months”.
-
District of Columbia paid leave does have an earnings-history requirement
What happened. This site described DC Paid Family Leave as having no eligibility threshold and being available from day one of employment. That was false. A claimant must have had covered DC wages reported in at least one of the five completed calendar quarters before the leave, so someone hired this quarter with no prior covered DC work cannot yet draw a paid benefit.
Who it affected. Overstated. A newly hired DC worker with no earlier covered wages was shown paid weeks she could not actually claim yet. DC's eligibility remains among the most permissive in the country — no tenure and no hours requirement — but it is not day-one universal.
Primary source. 7 DCMR § 3500.1, as published by the DC Department of Employment Services.
-
A California benefit that fell as earnings rose
What happened. California's schedule had been encoded as a straightforward 90% / 70% two-tier structure. The state's actual published table also contains flat bands that join those two curves. Omitting them turned a continuous schedule into a step function that ran backwards at the join.
Who it affected. Understated, and badly at one point. A claimant with $65,124 of income was shown $876.67 a week where the state pays $1,127 — $250 a week less for earning $4 more. The most common case on this site, an $80,000 earner, was underpaid by $50 a week for the length of her leave.
Primary source. The California Employment Development Department's published weekly benefit amount table, whose flat $1,127 band is the crossover where 90% of one bracket equals 70% of the next.
-
Colorado's 2027 maximum was published here as its 2026 maximum for about seven hours
What happened. Colorado's state average weekly wage changes on 1 July, but the benefit figures derived from it do not change until the following 1 January. Those two dates were conflated and next year's cap was applied as though it were current.
Who it affected. Overstated for roughly seven hours before being reverted. Every Colorado claimant whose benefit is limited by the cap was shown $66.57 a week more than the state actually pays.
Primary source. Colorado's own FAMLI guidance, which states that benefits “are capped at $1,381 per week as of January 1, 2026”, and its benefits calculator, which computes from the 2025–2026 average weekly wage. Archived copies confirm the cap held its previous value through December even after the July wage change.
-
The figure in Maine's benefit-cap field was not a cap at all
What happened. Maine's maximum weekly benefit had been recorded as $1,198. That number is Maine's 2025 average weekly wage, which the statute uses in a completely different place — the base-period earnings test for eligibility. It had been lifted into the benefit-cap slot. The actual maximum is the state average weekly wage itself, $1,250 for the benefit year beginning 1 July 2026.
Who it affected. Understated. Higher earners in Maine were shown a maximum about 4% below the real one.
Primary source. 26 M.R.S. § 850-C(3): “The maximum weekly benefit amount calculated under subsection 2 is the state average weekly wage.” Corroborated by Maine's official benefit estimator and the Workers' Compensation Board's published average weekly wage.
-
Oregon's superseded figures were taken from the wrong side of the state's own comparison table
What happened. Oregon publishes its annual change as a before-and-after table. The previous year's values were read from the left-hand column and encoded as current, so the minimum and maximum on this site were the ones Oregon had just replaced.
Who it affected. Understated for claimants at the maximum, whose cap should have been $1,692.16 rather than $1,636.56, and for those at the minimum. Oregon fixes these figures at the start of a benefit year, so anyone who began a claim on or after 28 June 2026 was affected.
Primary source. Oregon Employment Department announcement of 29 May 2026, listing “$1,636.56 -> $1,692.16 per week” for benefit years beginning on or after 28 June 2026.
-
Rhode Island's maximum was stale for two weeks after its July turnover
What happened. Rhode Island's maximum weekly benefit changes on 1 July on a benefit-year basis, not on 1 January. The site carried the previous year's $1,103 for about two weeks after the state moved to $1,150, because the monitoring was pointed at a fixed link to the prior year's announcement — an address that can never change, since the state publishes each year's release at a new one.
Who it affected. Understated by $47 a week for Rhode Island claimants at the maximum who began a benefit year on or after 1 July 2026. Rhode Island fixes the rate for the life of a claim, so an affected claim keeps the figure it started with.
Primary source. Rhode Island Department of Labor and Training's “2026 UI and TDI Quick Reference”, effective 1 July 2026.
-
Minnesota's maximum is not a calendar-year figure
What happened. Minnesota's maximum weekly benefit had been stored as a calendar-2026 value. It is keyed to a benefit period running from the last Sunday in October to the last Sunday in October, so $1,423 applies to leaves established between 26 October 2025 and 24 October 2026 — and a different maximum applies for the final weeks of calendar 2026.
Who it affected. No payment was wrong at the time of the correction, but the figure was scoped incorrectly and would have become wrong from 25 October 2026 onward. Minnesota fixes the amount when a claim is established, so anyone who begins leave on or before 24 October 2026 keeps this figure for the whole leave.
Primary source. Minn. Stat. § 268B.04, which provides that once established, an applicant's weekly benefit amount is not affected by the last-Sunday-in-October change.
-
New Jersey benefits were paid to the cent where the statute rounds down to a whole dollar
What happened. New Jersey's benefit rate is computed to the next lower multiple of one dollar after the replacement rate and cap are applied. The calculator was showing unrounded cents.
Who it affected. Overstated by less than a dollar a week — an $80,000 earner was shown $1,307.69 where the statute pays $1,307. Small, but it made the figure disagree with the state's own determination letter.
Primary source. N.J.S.A. 43:21-40(c): “Each individual's benefit rate shall be computed to the next lower multiple of $1.00 if not already a multiple thereof.”
How these get caught
Most of them came from a single deliberate exercise: re-deriving every figure from the issuing agency rather than checking it against the last thing we had recorded. That distinction is the whole game. A figure verified against your own notes will agree with your own notes.
The rest of the process is described on the methodology page — which sources are monitored, how often, why the audit calendar runs six times a year rather than annually, and why a figure an agency has not published yet is left blank here rather than estimated.