Federal Red Flag Orders

Create a federal court order that temporarily removes firearms from a person shown to be a danger to themselves or others, available in every state rather than only the twenty-two that have their own.

AI evaluation · not yet reviewed by a human

This evaluation was produced and sourced by an AI model; a human review is still pending. Figures and conclusions may still change. The review log is at the foot of the page.How review works →

Twenty-two states and the District of Columbia let a judge order the temporary removal of firearms from someone found to be a danger; in the other twenty-eight there is no such procedure, and the federal role since 2022 has been to fund state programmes rather than to require them. The bill before Congress creates the order in federal court: a family member or a police officer petitions, the court may issue an interim order on the petitioner's evidence alone, and a full hearing with the respondent present follows within fourteen days, after which an order lasts up to a year and may be renewed or dissolved. Firearms are surrendered for that period and returned when it ends. Making a false petition is itself an offence. This evaluation looks ten years ahead.

Balance

Much better for the future · 0.82 previous scale

Balance on the previous scale. The Bilanz 2.0 simulation is not yet available for this evaluation. The category comes from the share of the debate on the pro side (r).

For 38 · 82 % Against 8.7 · 18 %

Size class: small Scale of this evaluation: Normalised Impact — unitless, calibrated to this topic. For comparison: one point here is worth roughly 100 million euro per year. The whole positive side rests on one ratio: how many removal orders it takes to prevent one suicide. The best study, covering 4,583 cases across four states, puts it at one in seventeen to twenty-three, revised upward from an earlier estimate of one in ten. The conservative end of that range is used here, together with a deliberately modest guess at how many orders a federal law would actually produce in states that have chosen not to have one — eight thousand a year. Double either number and the measure is twice as strong; assume that courts in states hostile to the idea issue almost none, and it does almost nothing at all. How we score →

Arguments for

Arguments against

6 arguments evaluated · Scoring v1.3 Δ absolute +29.3

Arguments — For

2 arguments

People who do not die

33of 100

Firearm suicide is the largest category of American gun death and the one where the method decides the outcome: the same act with a different method is survived nine times out of ten. Removing the gun for a year removes the method during the crisis.

Value 9.5 · LifeImpact 5.4Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is people alive at the end of the period who would otherwise have died by their own hand, with a tenth added on the same anchor for the families, as this site does wherever it counts violent death. It sits just below the top of the scale rather than at it, because a proportion of what these orders prevent is an attempt rather than a death and the estimate used is of deaths averted. What is priced is the life, not the crisis, and not the treatment that may follow the order. Nothing is counted for the people who are removed from danger and then relapse, whose outcome the estimate already reflects. The value sits just below the maximum: the stream is life, reached by taking the method away during the weeks the danger is highest.

Swanson and colleagues examined 4,583 removal cases across California, Connecticut, Maryland and Washington and estimated that one suicide is prevented for every seventeen to twenty-three orders issued, and for every thirteen to eighteen where the record documents suicidality ^[4]. That is a revision upward of the earlier and widely quoted figure of one in ten. The conservative end, one in twenty-three, is used here. How many orders a federal law would add is the softer number: the twenty-eight states without their own programme have chosen not to have one, and their courts and police would be starting from nothing. Eight thousand additional orders a year is used, in a range from three thousand to twenty thousand — well below the volume the existing state programmes now produce. That gives about 348 deaths a year, or 536 million euro with the family surcharge. The Impact is the largest here and it is the product of two numbers, one of them measured and one of them a guess about how a reluctant state behaves.

▸ Show calculation ▾ Hide calculation

	Additional removal orders a year in the states without a programme Setting, range 3,000 to 20,000: the twenty-eight states concerned have chosen not to legislate, and their courts and police would be starting from nothing [3]	eight thousand	8,000 orders a year
÷	Orders needed to prevent one suicide the conservative end of the one-in-seventeen-to-twenty-three range measured across four state programmes [4]	23	348 deaths a year
×	Value of a life, with a tenth added for the families the value of a statistical life used across this site and the surcharge this site applies to violent death	1.4 million euro each, plus 10 percent	536 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	5.36

Score 5.36 Impact × 9.5 Value × 6.5 Plausibility ÷ 10 = 33 of 100

The counterfactual is people with the same documented risk profile who were not subject to an order, which the study constructs by comparing observed deaths among respondents against the deaths expected in a matched population at equivalent risk. The design is quasi-experimental in construction and observational in nature: nobody randomised who received an order, and the matching is on recorded risk rather than on anything exogenous. The confounder that matters is that a person who comes to a court's attention is in a different state from a statistically matched peer — usually worse, which would make the true effect larger than measured, occasionally better, because someone whose family cares enough to petition has support that others lack. The authors address it by reporting the estimate separately for cases with documented suicidality, where the effect is stronger, which is the direction the mechanism predicts. Reverse causation does not arise: the order follows the risk. Substitution to another method is inside the estimate, since the comparison is of deaths rather than of firearm deaths. The Plausibility is above the middle: the ratio is measured across four programmes and thousands of cases, on a comparison that matches rather than randomises.

evidence basis: Multiple precedents · P ceiling 7 identification: Controlled · rung ceiling 7

Counterfactual: a matched population at equivalent documented risk who were not subject to an order. Design: controlled — observed deaths among 4,583 respondents against expected deaths in a matched group, with no exogenous variation in who receives an order (Swanson and colleagues, 2024) [4]. Confounder: people who come to a court's attention differ from matched peers, usually towards higher risk, which would understate the effect, but occasionally towards more family support, which would overstate it; addressed by reporting the estimate separately for documented suicidality. Direction: reverse causation does not arise, the order follows the risk. Ceiling: controlled 7.0 binds below the multiple-precedents ceiling of 8.5.

Threats that are not carried out

5.3of 100

About a third of removal orders are sought because someone has threatened another person rather than themselves, and prosecutors in California have documented cases in which a threatened mass shooting did not happen after an order.

Value 9.5 · Life and healthImpact 1.2Plausibility 4.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is other people not killed or wounded: partners, colleagues, the crowd at an event. It sits just below the top of the scale in the class this site uses for life and protection from crime, with the family surcharge applied as for the suicides above. It is a separate argument because the mechanism is different — a suicide is prevented by removing a method during a crisis, a homicide by removing it from someone who has announced an intention against another person — and because the evidence for the two is of different quality. Nothing is counted for the fear the threat itself causes, which the order does not undo. Nothing is counted for the offender. The value sits just below the maximum, and this stands apart from the suicides because removing a method and disarming a threat are not the same mechanism.

Roughly a third of removal petitions in the states with programmes concern a threat to another person rather than to the petitioner's subject alone. One homicide prevented per hundred orders is used here, in a range from one in fifty to one in four hundred — a figure with no measured basis, unlike the suicide ratio. Applied to eight thousand additional orders, that is about eighty deaths a year, or 123 million euro with the family surcharge. The reason the ratio is set an order of magnitude below the suicide ratio is that the suicide effect is measured and this one is not: what exists is a case series from California in which prosecutors documented twenty-one instances of a threatened mass shooting followed by an order and no attack, which establishes that the mechanism can operate and cannot establish how often it does. Nothing is counted for non-fatal assaults prevented, which would enlarge this. The Impact is a fifth of the suicide argument, and the ratio behind it is the least evidenced number in this evaluation.

▸ Show calculation ▾ Hide calculation

	Additional removal orders a year [3]	from the argument above	8,000 orders a year
÷	Orders per homicide prevented Setting, range 50 to 400: an order of magnitude below the measured suicide ratio, because for homicide only a case series exists [5]	100	80 deaths a year
×	Value of a life, with a tenth added for the families the value of a statistical life used across this site and the surcharge this site applies to violent death	1.4 million euro each, plus 10 percent	123 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	1.23

Score 1.23 Impact × 9.5 Value × 4.5 Plausibility ÷ 10 = 5.3 of 100

The counterfactual is the same threatening individuals without an order, and nothing constructs it. The chain is named and complete: a threat is made, a court removes the firearms, the person does not have them during the period of highest risk. The evidence is a case series — twenty-one documented instances in California — which describes outcomes after orders without any comparison group, so it establishes possibility rather than frequency. The confounder that matters is that the same cases usually attract other interventions at the same time: arrest, prosecution, a protective order, a mental health hold, any of which could account for the absence of an attack. That is named and unresolved. Reverse causation does not arise. Since the mechanism is well described and nothing measures its size, the number rests on a complete chain rather than a finding. The Plausibility is below the middle: the mechanism is documented case by case and its frequency has never been estimated against a comparison group.

evidence basis: Mechanism · P ceiling 6 identification: Mechanistic · rung ceiling 6 band: Chain closed, unevidenced · P 4–5

Counterfactual: the same threatening individuals without an order — not constructed by any study. Design: mechanistic — a California case series of twenty-one documented instances with no comparison group establishes possibility, not frequency [5]. Confounder: arrest, prosecution, protective orders and mental health holds usually accompany the same cases and could account for the outcome; named and unresolved. Direction: no reverse causation. Ceiling: mechanistic 6.0 binds. Band: chain closed but unevidenced — every link named, the co-intervention counter-mechanism stated, only the frequency missing.

Nothing measured argues against the claim; what is missing is any estimate of how often an order prevents a homicide, against a comparison group. The counter-mechanism — that arrest and prosecution usually accompany the same cases — is named and unresolved. Read back: about half the time, orders of this kind prevent roughly one homicide per hundred issued.

Open: The states with programmes hold petition records with the reason for each order, and death records are linked by state. Following respondents in threat-to-others cases against matched non-respondents, as has been done for suicide, would give the first estimate and could carry this above 6.

Arguments — Against

4 arguments · top 3 shown

Owning under a condition

4.1of 100

In the twenty-eight states with no such law, forty-five million gun owners would for the first time hold their firearms subject to a procedure a relative or an officer can start against them.

Value 9 · Rights and libertyImpact 0.9Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is a right held under a new condition: not taken from anybody in particular, but exercised from now on in the knowledge that a family member or a police officer can begin a process to suspend it. This site places that at the constitutional core, and prices it as a restriction that applies broadly and bites lightly, because for almost everyone nothing happens. It is separate from the argument above, which prices what happens to the people an order actually reaches; a rule that touches forty-five million people slightly and eight thousand heavily is two facts, and averaging them would hide both. Nothing is counted for the disapproval of people who dislike the law without owning a firearm. The value is at the constitutional core, and it is priced at the light end because for almost everyone the condition is never triggered.

About forty-five million adults own a firearm in the twenty-eight states without a removal law. Two euro per owner per year is used, in a range from 0.5 to 8 — the band this site applies to a restriction that reaches a large population and bites lightly, the same one used for the loss of anonymity in the platform evaluations elsewhere on this site. That gives 90 million euro a year, more than the argument above about the people actually disarmed, which is what happens when a small burden is spread over forty-five million people. The figure would be larger if the procedure were used more than assumed, and the number of orders is set low; the two move in opposite directions, since a law nobody uses restricts nobody in practice. Nothing is counted for owners in the twenty-two states that already have such a law, for whom the federal procedure adds little. The Impact is four times the burden on the people an order reaches, purely because there are five thousand times as many of them.

▸ Show calculation ▾ Hide calculation

	Firearm owners in the states without a removal law [3]		45 million adults
×	Value of the condition per owner per year Setting, range 0.5 to 8 euro: the band this site uses for a restriction that reaches a large population and bites lightly	2 euro	90 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	0.9

Score 0.9 Impact × 9 Value × 5 Plausibility ÷ 10 = 4.1 of 100

The condition is definitional: the statute creates a procedure and it applies to everyone who owns a firearm in the states concerned. The counterfactual is those states as they are, with no such procedure. What is a convention rather than a measurement is the two euro, which is this site's band for a broad light restriction and not anything established about American gun owners. The confounder is the one named above and it cuts both ways: if courts in these states issue very few orders, the restriction is nominal and this argument shrinks with the ones opposite; if petitions are frequent, both grow. Reverse causation does not arise. Nothing measures what a conditional right is worth to the person holding it, so the plausibility is held at the top of what a valuation of this kind can carry. The Plausibility is at the middle: the condition is certain and what it is worth to those it applies to is a convention.

evidence basis: Plausibility · P ceiling 5 identification: Definitional · no rung ceiling

Counterfactual: the twenty-eight states as they are, with no removal procedure. Design: definitional — the statute creates the procedure and it applies to every firearm owner in those states; no behavioural link carries the quantity. Confounder: how often courts actually issue orders, which moves this argument and the ones opposite in the same direction; named. Direction: not applicable. Ceiling: plausibility 5.0 binds, because the per-owner figure is this site's convention for a broad light restriction and not a measurement.

States that decided otherwise

2.7of 100

Twenty-eight legislatures have considered removal laws and declined to pass them, several more than once. A federal order applies in those states regardless of that decision.

Value 6 · State discretionImpact 0.9Plausibility 5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is a decision taken away from the level that took it: twenty-eight legislatures that examined this question and answered no, overridden by a federal procedure that reaches into their courts. This site places that with the working order of the political system rather than at the constitutional core, because what is at stake is which body decides rather than whether a right exists. It is separate from the two arguments above, which price what the rule does to individuals; this prices what it does to the arrangement by which such rules are made. Nothing is counted for the states that already have such a law, whose choice is unaffected. Nothing is counted for the litigation the arrangement would attract, which is an administrative cost and small. The value sits in the middle-upper part of the scale, where this site places the working order of the political system, rather than at the constitutional core.

About 180 million people live in the twenty-eight states without a removal law, and their legislatures have in most cases considered and rejected one. Fifty cents per resident per year is used, in a range from 0.1 to 3 euro — well below the seventy euro this site uses where a direct popular vote is set aside, because a bill that fails in a legislature is a weaker expression of a settled choice than a referendum, and well above zero because the choice was made repeatedly and deliberately. That gives 90 million euro a year. The figure would be smaller if the federal procedure ran alongside state law rather than through federal courts, and larger if it displaced state procedures where they exist, which it does not. The Impact is the same size as the burden on gun owners in those states, which is what happens when a small per-person figure meets a very large population.

▸ Show calculation ▾ Hide calculation

	People living in states that have declined to create a removal law [3]		180 million residents
×	Value of the decision being overridden, per resident per year Setting, range 0.10 to 3 euro: far below the seventy euro this site uses where a direct popular vote is set aside, because a bill that fails in a legislature is a weaker expression of a settled choice	0.50 euro	90 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	0.9

Score 0.9 Impact × 6 Value × 5 Plausibility ÷ 10 = 2.7 of 100

The fact is not in dispute: a federal procedure applies in states that declined to create their own. The counterfactual is those states as they are. What is a convention rather than a measurement is the fifty cents, and it is the softest number in this evaluation — this site has one anchor for a set-aside popular vote and none for a legislature's decision not to act, so the figure is derived by analogy and by judgement about how much weaker the second is than the first. The confounder that matters is that a legislature's failure to pass a bill is not always a decision against it: bills die for reasons of time, procedure and committee assignment, and treating every failure as a settled popular choice overstates the harm. That is named and unresolved. Reverse causation does not arise. The Plausibility is at the middle: the override is certain and the price put on it is derived by analogy from a single anchor for a different thing.

evidence basis: Plausibility · P ceiling 5 identification: Definitional · no rung ceiling

Counterfactual: the twenty-eight states as they are, having declined to legislate. Design: definitional — the federal procedure applies regardless of state law; no behavioural link carries the quantity. Confounder: a bill failing in a legislature is not always a decision against it, since bills die for reasons of time and procedure, which would overstate the harm; named and unresolved. Direction: not applicable. Ceiling: plausibility 5.0 binds, because the per-resident figure is derived by analogy from this site's anchor for a set-aside popular vote and is not a measurement.

Disarmed before the hearing

1.5of 100

An interim order can issue on the petitioner's evidence alone, with the respondent's own hearing up to fourteen days later. For that fortnight a person keeps none of their firearms and has said nothing to the court.

Value 9 · Rights and libertyImpact 0.2Plausibility 7

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the respondent's position: a right suspended on one side's account, property taken, and a court record naming them a danger before they have answered. This site places that at the constitutional core, because what is at stake is a right removed by the state without the person being heard first. What is priced is the deprivation itself and the finding that accompanies it, spread across everyone an order touches rather than only those later found not to meet the standard, since the interim period applies to all of them. Nothing is counted for the stigma that persists after an order is dissolved, which is real and is not measured by anything here. Nothing is counted for the property itself, which is returned. The value is at the constitutional core because a right is suspended before the person is heard, and the amount per person is what keeps the argument small.

Eight thousand additional orders a year is the same figure used on the other side. What each costs the respondent is put at 3,000 euro, in a range from 1,000 to 20,000 — this site's band for an unfounded suspicion recorded against a named person, applied here to everyone an order touches rather than only to those later found not to meet the standard. In the state programmes roughly a fifth to a third of interim orders are not confirmed at the full hearing, and those cases sit at the upper end of the band while a confirmed order sits near the lower, because the person was in fact found dangerous. That gives 24 million euro a year. Nothing is added for the cases in which firearms are not returned promptly after an order lapses, which the state programmes report as a recurring administrative failure and which nothing here measures. The Impact is a twentieth of the lives on the other side, which is the arithmetic of pricing a fortnight without firearms against a death.

▸ Show calculation ▾ Hide calculation

	Additional removal orders a year [3]	the same figure used on the other side	8,000 orders a year
×	Cost to the respondent of an order Setting, range 1,000 to 20,000 euro: this site's band for an unfounded suspicion recorded against a named person, applied to every respondent because the interim period applies to all of them	3,000 euro	24 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	0.24

Score 0.24 Impact × 9 Value × 7 Plausibility ÷ 10 = 1.5 of 100

The procedure is written into the bill, so what happens to a respondent is not in question: an interim order, a surrender, a hearing within fourteen days. The counterfactual is the twenty-eight states where no such procedure exists. The chain has no behavioural link carrying the quantity — the number of respondents is the number of orders, which is set on the other side and used identically here. What is a valuation rather than a finding is the amount per respondent, and it is taken from this site's own band rather than measured for this situation. The confounder that would raise the figure is the share of interim orders not confirmed, which varies widely between state programmes and is the reason the band is wide. Reverse causation does not arise. The Plausibility is high: the procedure is statutory, the number of people is the same one used opposite, and only what it is worth to each is a convention.

evidence basis: Precedent · P ceiling 8 identification: Definitional · no rung ceiling

Counterfactual: the twenty-eight states where no such procedure exists. Design: definitional — the interim order, the surrender and the fourteen-day hearing are written into the bill, and the number of respondents equals the number of orders used opposite. Confounder: the share of interim orders not confirmed at the full hearing varies widely between state programmes and sets where in the band a case sits; named. Direction: not applicable. Ceiling: precedent 8.5, held at 8.0 because the per-person amount is this site's convention rather than a measurement.

Somebody has to serve them

0.4of 100

Every order means a petition, a hearing, an officer at a door and a store for the firearms until the year is up. Serving removal orders is also among the more dangerous things an officer is asked to do.

Value 5 · Enforcement costImpact 0.1Plausibility 6.5

▸ Show reasoning & sources ▾ Hide reasoning & sources

The stream is the court and police time an order consumes: the petition, the interim decision, the full hearing, the service, the storage and the return. It is priced at the middle of the scale, like other money spent on running a rule. The danger to the officers serving the orders is inside this figure rather than beside it, because the number of deaths involved is a fraction of one a year at this volume and pricing it separately would give it a precision it does not have. Nothing is counted for the respondents' own legal costs, which are inside the argument about what an order costs them. The value is the middle of the scale, the level this site uses for the cost of running a rule.

Eight thousand additional orders a year at about 1,500 euro each — a petition, an interim decision, a hearing with both parties, service by officers, and storage for up to a year — comes to 12 million euro a year at the ordinary weight for public money. The per-order figure is taken from what the existing state programmes report spending, which varies by a factor of three between states with dedicated units and states where the work falls on ordinary patrol. Inside the same figure sits the risk to the officers who serve the orders: at the rates the state programmes have experienced, eight thousand services a year produce well under one death, which at this site's value for a life is a few hundred thousand euro and is left inside the total rather than given a line of its own. Nothing is added for the storage facilities themselves, which the state programmes report as their most persistent shortage and which would raise the figure where none exist. Nothing is deducted for the police time that would have gone into the same cases anyway, since most of these people are already known to somebody. The Impact is the smallest in this evaluation, about a fiftieth of the lives it buys.

▸ Show calculation ▾ Hide calculation

	Additional removal orders a year [3]	the same figure used on the other side	8,000 orders a year
×	Court, police and storage cost per order what the operating state programmes report spending, which varies by a factor of three between dedicated units and ordinary patrol [3]	1,500 euro	12 million euro a year
÷	Normalised Impact scale of this evaluation	100 million euro a point	0.12

Score 0.12 Impact × 5 Value × 6.5 Plausibility ÷ 10 = 0.4 of 100

The costs are observed rather than predicted. The counterfactual is the same courts and police without the procedure. The state programmes publish what they spend and how many orders they process, so the per-order figure is a precedent transfer rather than a projection; what is uncertain is which end of the observed range a new federal procedure would sit at, and it would probably start at the expensive end because no dedicated units exist. That is named. The chain has no behavioural link carrying the quantity beyond the number of orders itself, which is set on the other side and used identically. Reverse causation does not arise. The Plausibility is above the middle: the per-order cost is observed in operating programmes and the transfer to a new federal procedure is the only uncertainty.

evidence basis: Multiple precedents · P ceiling 7 identification: Definitional · no rung ceiling

Counterfactual: the same courts and police without the procedure. Design: definitional — the number of orders determines the number of proceedings, and the per-order cost is observed in operating programmes. Confounder: a new federal procedure would start at the expensive end of the observed range because no dedicated units exist; named. Direction: not applicable. Ceiling: multiple precedents 8.5, held at 7.0 for the transfer from established state units to a new federal procedure.

Summary

This comes out strongly positive and it does so for a reason worth stating plainly: the benefit is measured in lives and the costs are measured in fortnights without a firearm and fifty cents a head for a decision taken at the wrong level of government. At the conservative end of the best available estimate — one suicide prevented for every twenty-three orders — eight thousand additional orders a year prevent about 348 deaths. The objections are real and they are small on this scale: an interim order issues before the respondent is heard, forty-five million owners in the affected states would hold their firearms under a new condition, and twenty-eight legislatures that considered such a law and declined would be overridden. What could change the result is not any of those but the volume: a federal procedure that courts in reluctant states barely use would prevent few deaths and impose few costs, and this evaluation assumes eight thousand orders where the existing state programmes now produce more.

Outlook — effect over time

Much better for the future · 0.82 previous scale

Sources

GovTrack: H.R. 7599, Federal Extreme Risk Protection Order Act of 2026. govtrack.us
Johns Hopkins Center for Gun Violence Solutions: Gun violence in the United States 2023. publichealth.jhu.edu
Johns Hopkins Bloomberg School of Public Health: Research on extreme risk protection orders. publichealth.jhu.edu
Swanson and colleagues, Journal of the American Academy of Psychiatry and the Law: Suicide prevention effects of extreme risk protection order laws in four states. jaapl.org
Wintemute and colleagues, Annals of Internal Medicine: Extreme risk protection orders intended to prevent mass shootings: a case series. acpjournals.org

Last reviewed by Claude Opus 5 · September 6, 2026 · 2× AI, not yet reviewed by a human

September 6, 2026AI reviewClaude Opus 5record updated
i_spanne an allen 6 Argumenten, normalisierung an 3 von 6 (leben mit Angehoerigen-Zuschlag, eur-nennwert) — die drei Grundrechts- und Foederalismus-Setzungen haben keinen Anker und bleiben ohne Block, kein de-Anker gebraucht. eintritt_gruppe orders-actually-issued, kein Transfer im Record. Kategorie kippt von Deutlich besser (r 0,82) auf Ausgeglichen (P(D>0) 0,64) — vier Con-Karten sind definitorisch, nur die Pro-Seite zieht ein Eintrittslos.
September 6, 2026AI reviewClaude Opus 5First evaluation
First evaluation: the conservative end of the one-in-seventeen-to-twenty-three suicide ratio, with a deliberately low guess at order volume.

Evaluations are produced with AI support and reviewed on a schedule for new developments; human passes are marked separately.How we review →