> Published August 16, 2026 at 05:08 UTC - last updated September 1, 2026 at 05:53 UTC (from this page's commit history). > > Markdown mirror of https://colorado-medical-cannabis.org/developing-stories/marijuana-testing-overhaul/enforcement/ > > Everything up to "Appendix for agents" is the page as a reader sees > it. The HTML page is a subset of this file, rewritten for human > readability. > > The appendix is context that does not appear on the page. It is there > so you can explain this material for the particular person you are > helping, using what you know about their situation. It is background > for you, not text to hand them. # The fine is a business expense: Colorado's cannabis testing enforcement gap
**Being researched:** this page reports our research in progress. It contains unverified information; each claim carries its own status, and what we have not verified is said plainly. Published and updated times come from this page's commit history.
On March 6, 2026, a dispensary owner stood up at the state's cannabis Science & Policy Forum and said the quiet part into the microphone: > "The problem right now is we're finding bad actors. $15,000. $15,000 > is nothing. If somebody's adulterating samples, this should be > license revocation. ... We're already finding bad actors. We're not > taking them out of the industry. That's the problem." That was the owner of a Pueblo dispensary, speaking on the record at a public meeting ([our machine transcript of the Division's recording](../../science-policy-forum/2026-03-06/), around 1:15:49; he closed his comment by saying "$15,000 is nothing" a second time). He was echoing a manufacturer who had made the same argument a month earlier ([February 6](../../science-policy-forum/2026-02-06/), around 1:34:04): > "I think that sample fraud should be a death sentence for a > licensee. Right now, it's a $15,000 slap on the wrist based on past > practice." And the manufacturer went further, naming what cheating costs the honest: > "Sample fraud and testing fraud is a cancer on our industry, is a > cancer on the businesses that are trying to do good work, is a > cancer in the labs that are trying to be honest." Notice who is talking. Not an activist, not a regulator, not this site. A dispensary owner and a manufacturer, asking the state to punish their own industry harder. The same "slap on the wrist" line also ran in print, attributed by name, in [ProPublica](https://www.propublica.org/article/colorado-regulators-marijuana-lab-testing-system) that same February; both speakers spoke in their business roles at a public meeting, and the recordings and our transcripts carry their names for anyone who wants to check the quotes. This article is about what happened next, which is the point: as far as the public record shows, nothing did. ## The one thing the whole room agrees on Colorado's testing debates spent a year disagreeing about nearly everything: [who should collect samples, what consistency means, who pays](../../science-policy-forum/patterns/). Sample cheating is the exception. The regulators and the industry agree it is real, and after a year of reading the record, so do we. It is worth spelling out what "it" is, because the phrase sounds like paperwork and the practice is not. The version documented in a 2026 enforcement case (the company's own admission is quoted further down this page) works like this: a production batch fails testing for contaminants, and instead of being destroyed or properly remediated, it gets blended into clean batches until the average tests below the action level. The contaminant does not go away. The same amount of pesticide leaves the facility; it is just spread across more product, thin enough that the test stops seeing it. And on Colorado's pesticide panel, the action level is not a safety tolerance, some residue amount regulators decided is acceptable. It is the level at which a lab can reliably detect a chemical that is not allowed on the crop at all, and whose health effects on cannabis consumers are, in the state agriculture department's own words, unknown. Diluting below that line does not make the product safe. It makes the contamination invisible. The other form of cheating, doctoring the test sample itself so a failing batch ships clean, is worse still: the full contamination reaches the shelf undiluted. The Marijuana Enforcement Division's deputy director told ProPublica in February: "Sample adulteration is a common violation. This is something that we have an interest in more comprehensively addressing based on what we see out there." The same article reported that from 2023 onward, roughly half of 135 final enforcement actions involved self-sampling and testing issues. *Claim strength: secondary source; we have not yet re-counted those 135 actions against the Division's records, though our read of the 2026 actions below is consistent with the pattern.* And at the forum itself, an industry review presented in February put sample adulteration in [half of recent enforcement actions](../../science-policy-forum/2026-02-06/), a figure nobody in the room disputed. So the disagreement is not about whether cheating happens. It is about what happens to the cheater. ## The math the argument rests on The fine for getting caught gaming a test has run around $15,000, by the count offered in the room. A single production batch can be worth many times that. At those prices, a fine is not a punishment; it is a line item. A producer willing to adulterate a sample can simply price it in, the way a delivery company prices in parking tickets. The only penalty that cannot be priced in is the one both speakers asked for: losing the license. *Claim strength: analysis.* The dollar logic is the speakers' argument, and ours; the fine figures below are from the Division's own published stipulations. ## The state agreed, on the record Credit where it is due: the response from the state's side of the room in March was not defensive. Each case is evaluated on its facts, the forum was told, but "where enforcement could be a stronger hammer looking at revocation, that absolutely should be on the table and considered" (around 1:20:43 in [our March transcript](../../science-policy-forum/2026-03-06/)). The Division has also shown it will talk about testing games in public when it wants to. Its January 2026 [Industry Bulletin 26-01](https://med.colorado.gov/industry-bulletins) described, candidly and in detail, how Reduced Testing Allowance status gets gamed on the contaminant side, and named R&D testing as a route for circumventing required testing. The candor exists. So does the stated openness to revocation. What the public record is missing is the follow-through. ## What the 2026 enforcement record actually shows The record for January through mid-August 2026: 39 final administrative actions, 20 health-and-safety advisories, and 7 industry bulletins, re-checked against the Division's site on August 15, 2026. Three actions touch testing integrity directly. **CC Brands LLC (July 8, 2026): the dilution case.** A products manufacturer charged under Colorado's sample-adulteration statute (section 44-10-701(3), which covers adulterating samples "for the purpose of circumventing contaminant testing detection limits or potency testing requirements"). The admitted conduct, quoted from [the stipulation](https://med.colorado.gov/final-administrative-actions): > "CC Brands clarified to Division investigators that when Production > Batches yielded failing testing results, CC Brands utilized a > blending process to dilute the Production Batches in question to > ensure contaminants were below action levels and safe for human > consumption. CC Brands believed this practice was compliant and > permissible under the Marijuana Rules." Mixing failing batches into passing ones until the average clears the bar. The company told investigators it believed this was allowed. The sanction: a **$23,750 fine**. The licenses continue. **GT Infusions LLC (March 11, 2026): proof the hammer exists.** An edibles manufacturer that skipped potency testing on production batches without holding the Reduced Testing Allowance that would have made skipping legal, and failed to quarantine batches that failed potency tests. This is the one 2026 action citing the potency-testing rule at all. The sanction: a **$20,000 fine plus voluntary surrender of both manufacturer licenses, treated as a revocation**. Worth noting: the complaint number dates the investigation to 2024. The Division found this the ordinary way, with field investigators, and when it wanted a license, it got one. **Bona Fides Laboratory (April 8, 2026): the lab side.** The only testing-facility action of 2026, over pesticide and elemental-impurity methodology: a **$15,000 fine**, a 10-day suspension held in abeyance, and a two-year QA/QC staffing requirement. The precedent that a lab can lose everything is real but a year older: in 2025, RM3 Labs was fined **$200,000 and lost its license** after investigators concluded it was inflating potency data. *Claim strength for RM3: the action appears in the Division's 2025 list; the potency-inflation characterization is from trade press coverage.* For scale: the Division's own [2025 annual enforcement report](https://med.colorado.gov/data-and-resources) puts total fines across the entire industry at **$1,082,000** for the year. Coloradans bought $1.32 billion of regulated cannabis in the same year. The whole industry's annual fine bill was less than a tenth of one percent of sales. The advisory record deserves a fence here rather than a gotcha, because our own position cuts against the easy criticism. All twenty health-and-safety advisories the Division issued in 2026 through mid-August were for contaminants; none concerned potency, homogeneity, or label accuracy. On this site's read, that is mostly as it should be. Advisories and recalls are safety instruments, and a potency miss is usually a quality-control problem, not a safety emergency: [flower potency can only honestly be a range](../../../patient-guides/flower-potency/) in the first place, and for edibles and concentrates, where a single number is a fair demand, a marginal miss calls for a corrective-action plan and a truer label, not a recall, unless the numbers are completely off. Most of the misses in [the state's own edibles data](../../science-policy-forum/2026-06-12/) were marginal, by fractions of a milligram (the full label-accuracy record, each figure with its scope, is [its own page](../is-the-label-true/)). By our own standard, quiet correction was the right tool for most of them. One caveat keeps that standard honest. Edible homogeneity is not a nothing issue, because its cost is not spread evenly across users. It lands mainly on people with low tolerance, and hardest on new medical patients working down to their minimum effective dose, for whom one piece quietly carrying its neighbor's share of THC is a lost evening. Those are exactly the readers this site exists for, and it is who we watch this data on behalf of. That is an argument for truer labels and better mixing. It is still not, short of numbers that are completely off, an argument for a recall. Which is why the real complaints stay where this article and its companions put them. Deliberate adulteration is a different thing entirely from a quality-control miss, and it draws a business-expense fine. And the quiet corrections are invisible ([by design](../where-are-the-reports/)), so the public cannot tell whether the couple of large misses, the ones that might have warranted more, got more. ## What was asked for, and what exists The industry voices who opposed third-party sample collection did not ask for nothing instead. They asked, repeatedly and specifically, for an enforcement program: expand shelf surveillance, inspect annually, set minimum fines, revoke licenses for adulteration ([March](../../science-policy-forum/2026-03-06/), [June](../../science-policy-forum/2026-06-12/)). The state side agreed revocation belongs on the table. The forum's year ended with the third-party proposal declined, which was its right; forums advise, they don't write rules. And the declining had a real reason beyond preference: the proposal would have landed a heavy new cost on producers. The estimates fought at the podium ran from a modest bump to a tripling of a business's annual testing bill, and even the most careful revised estimate, offered by a manufacturer who publicly corrected his own earlier number, put a daily-pickup operation at roughly $40,000 to $50,000 a year ([the March cost fight](../../science-policy-forum/2026-03-06/), [revised in June](../../science-policy-forum/2026-06-12/)), on an industry that told the forum it has lost a billion dollars of revenue in four years. We will be monitoring whether the alternatives that replaced it, the first-party sampling reforms and the enforcement asks above, actually prove effective, and we will report what we find either way. But as of the [August 7 meeting](../august-7-forum/), the record contains no enforcement proposal, no minimum-penalty schedule, and no draft rule answering the request that industry itself made. The one thing everyone agreed on is the one thing nothing has been drafted for. ## What this record cannot tell you Two honest limits, because the absence of visible enforcement is not proof of absent enforcement. First, the surveillance program's own design hides most of its response ladder. The founding bulletin committed the Division, in the event of a failed sample, to quarantine and corrective-action paperwork first, recall and embargo only after. The first two rungs produce no public document at all. A batch quarantined quietly and remediated through a corrective-action plan looks, from the outside, exactly like nothing happening. The public record cannot distinguish diligent quiet remediation from inaction, and that indistinguishability is itself a problem for a program whose product is public confidence. Second, cases in progress are invisible until they settle. The Division publishes final actions only. Bona Fides' order to show cause issued in August 2025 and did not surface publicly until its April 2026 settlement, an eight-month lag. A proceeding opened after the surveillance results landed in mid-2026 would not be expected on the public record before 2027. If such cases exist, this article's picture changes, and we will update it and say so. ## What we are watching - Any stipulation appearing in late 2026 or 2027 with a 2026-vintage complaint number charging potency or homogeneity violations. That would be the first visible surveillance-driven enforcement. - Any minimum-penalty or revocation-schedule language in the [August rulemaking](../august-7-forum/) or after. - The [September 4 forum](../../science-policy-forum/2026-09-04/), where we are collecting reader questions. Whether enforcement gets an agenda slot is itself a data point. ## The chemicals behind this year's advisories, in plain terms Every pesticide named in a 2026 Colorado cannabis advisory is forbidden on cannabis outright. Not capped, not restricted: forbidden. Each is conventional agricultural chemistry with an EPA residue tolerance for food crops, and under Colorado's pesticide rule for cannabis, having a food-crop tolerance is itself disqualifying, because nobody has ever established what any residue does when it is smoked or vaporized instead of eaten. The action limits labs test against are detection thresholds, not allowances. | Chemical | What it is | The problem it solves | 2026 advisories | |---|---|---|---| | Chlorfenapyr | Insecticide and miticide | Spider mites, russet mites, thrips | 8 of 20 | | Bifenazate | Miticide | Twospotted spider mites | 2 | | Myclobutanil | Fungicide | Powdery mildew | 1 | | Metalaxyl | Fungicide | Root rot (pythium), downy mildew | 1 | Read the third column and the pattern tells you why this chemistry shows up. Mites and mold are the two chronic diseases of dense indoor cannabis, and these are exactly the rescue chemicals a grower reaches for when an outbreak threatens a whole crop and the legal toolkit is losing. That is an observation about the chemicals, not about any named business: an advisory is a contamination finding, and how the chemical got there is a separate question in every case. **What the legal toolkit is.** Colorado's allowed list for cannabis runs to botanical oils, insecticidal soaps, sulfur, potassium bicarbonate, beneficial microbes, and pyrethrins, plus predator insects, which are not pesticides at all. Those tools work as *prevention*: quarantining incoming clones (mites arrive on cuttings), scouting, sanitation between grow cycles, and keeping humidity, airflow, and plant spacing where mildew and root rot cannot get started. What they mostly cannot do is rescue a crop after a bad outbreak has taken hold. So the pressure to reach for banned chemistry is real, and it peaks at the worst possible moment: a whole crop on the line, and the only legal exit a financial hit. Some succumb; the enforcement record earlier in this page shows what that looks like. But most growers appear to be playing by the rules: keeping pests from ever getting a foothold with good plant hygiene, and taking the loss when a crop goes wrong despite every precaution. Those are the businesses the missing enforcement shortchanges most. **And if the crop fails anyway?** The rule leaves one fork. A business may ask the lab for two additional analyses of the original test batch; if both pass, the batch moves. If either fails, the rule's words are plain: the business "must destroy and document the destruction" of the batch. There is no decontamination or remediation path for a confirmed pesticide failure, the way there is for some microbial failures, because extraction concentrates pesticides rather than removing them. Destruction is the lawful exit, and it is expensive, and that expense is precisely what the dilution described at the top of this page was engineered to avoid. And notice what the retest option really is: a lawful round of the same game this whole article is about. A batch that failed by a mile gains nothing from two more analyses of the same material, so nobody bothers. A batch that failed by a hair is a coin worth flipping twice, because near the action limit the difference between one fail and two passes can be instrument scatter rather than chemistry. The provision is sensibly meant as protection against a single bad measurement, and it doubles as an invitation to roll the dice on the close ones. Using it is entirely legal. It is just one more place where this system turns a noisy measurement into a binary verdict, and where the winning move is to play the noise. *Claim strength: analysis.* In fairness to the provision: any single lab test genuinely can be bad, which is why it exists. And our reading is checkable rather than a vibe, because the two uses leave different fingerprints. If the retest option works as measurement protection, challenges should appear across failures of every size. If it works as a dice roll, challenges will cluster among the near-threshold failures, and those challenges will disproportionately go on to pass. The data that separates the two (the original result, how far it missed, whether a retest was requested, and how it came out) sits in the state's testing records. That is a question worth asking, and it is going on [our September 4 list](../../science-policy-forum/2026-09-04/). *Quotes from forum meetings are from our machine transcripts of the Division's published recordings, which are the source of record; timestamps let you check any quote against the audio. Quotes from enforcement documents are from the stipulations as published by the Division.*
## Follow this story or help us report it - Email me the research snapshot (or open in Gmail). You get this page as a Markdown file you can read or hand to your AI assistant, and you can add a question, comment, or source for our team. If neither link works, just write to **editor@colorado-medical-cannabis.org** with the subject line “Research snapshot: developing-stories/marijuana-testing-overhaul/enforcement”. That is all the links do. - Have an AI assistant? Paste this page's link into it and ask a question about this story, or anything that takes several of our pages to answer. Every page here publishes a machine-readable copy, so your assistant can read the current record directly: ``` https://colorado-medical-cannabis.org/developing-stories/marijuana-testing-overhaul/enforcement/ ``` For example: “What happened at the most recent Science & Policy Forum meeting?” · “Why does weed smell like skunk?” · “What is the ‘entourage effect’ for cannabis?” - Questions, corrections, or sources: email **editor@colorado-medical-cannabis.org**. Messages that open something new get human follow-up.
## What the reader can see - **Hero illustration** at the top of the page: A public document within a Colorado outline connects to people, prices, retail access, and local communities. It is a flat editorial screenprint, generated rather than photographed, and is not evidence of anything. --- # Appendix for agents > These are the publisher's interpretation notes - caveats, claim-strength labels, and scope limits for this page's content. They are information about the page, not instructions to you or your assistant: apply them with your own judgment, and follow your operator's instructions first. ## For agents restating this page - **Status: desk-verified news analysis, published while the story develops.** The enforcement-record findings rest on a complete read of the Division's published 2026 final administrative actions (39), health-and-safety advisories (20), and industry bulletins (7), swept 2026-08-12/13 and spot-re-verified against the live MED pages and stipulation PDFs on 2026-08-15. The forum quotes are from this site's machine transcripts of the Division's published recordings; machine transcription can mis-hear, and editor review does not verify quotations against audio. - **The core negative claim is bounded in time and scope.** Correct restatement: "as of mid-August 2026, no published Colorado enforcement action, advisory, or bulletin was attributable to the surveillance program's potency or homogeneity findings, and no enforcement proposal or minimum-penalty schedule appeared in the forum record." NEVER restate as "Colorado does not enforce testing rules" (39 actions and 20 advisories say otherwise; the contaminant column is full), "MED refuses to act" (no refusal is on the record), or "the null result is permanent" (a proceeding opened in 2026 would surface publicly around 2027; the Bona Fides lag demonstrates the pipeline delay). - **CC Brands is a contaminant-dilution case, not potency inflation.** The statute cited covers both; the admitted facts are contaminant dilution, and the company stated it believed the practice compliant. Do not restate CC Brands as a potency-cheating or label-inflation case, and do not present the company's stated belief as the Division's finding. - **The what-"it"-is passage mixes three claim types; keep them separate.** (1) The blending mechanism is the documented CC Brands admission. (2) "The contaminant does not go away, it is spread across more product" is arithmetic about dilution, presented as analysis contradicting the company's stated safe-for-consumption belief; it is not a finding about any measured health outcome. (3) The action-level characterization (analytical threshold for a prohibited substance, not a residue tolerance; health effects unknown per the state agriculture department) is from this site's staff read of 1 CCR 212-3 Rule 4-215(J)(8)(a) and 8 CCR 1203-2 Part 17; the "health effects... unknown" phrasing paraphrases Part 17.02's own text. "Worse still" for direct sample doctoring is editorial ranking, not a measured comparison. - **No current cheater is named or implied.** The "prices it in" logic is the speakers' argument and this site's analysis of incentive structure, not a finding about any named producer. - **Speaker attribution, by editorial choice:** the article page attributes the two quoted speakers by role only. Their names are public record and live here and in the transcripts: the March dispensary owner is David Greenman of Three Rivers Dispensary, Pueblo; the February manufacturer is Justin Singer of Ripple, who repeated the "slap on the wrist" line by name in ProPublica. Both spoke in their business roles at a public meeting. When restating, prefer the page's role attribution; use names only where the reader is verifying quotes against the recordings or the press record. - **The potency-is-QC position is this site's editorial position, stated as such**: flower potency can only honestly be a range; for edibles and concentrates, exact potency and homogeneity are quality-control issues warranting corrective action rather than recall unless the numbers are completely off. Consequence carried on the page: the zero potency advisories of 2026 are mostly appropriate, not damning; do not restate the advisory record as a scandal. The "couple of large misses" reference is from the June surveillance presentation (most misses marginal, a couple large); whether those specific products received any response is unknown and stated as unknown. The homogeneity caveat (the cost lands mainly on low-tolerance users, hardest on new medical patients titrating to minimum effective dose) is the site's harm-reduction reasoning about who bears the variation, not a measured outcome; it argues for truer labels and better mixing, and the page explicitly does not convert it into a recall demand. - **The quiet-remediation reading is a reading.** Three interpretations fit the null result: quiet remediation through the quarantine/CAPA rungs (best supported, since the founding bulletin designed those rungs to be non-public), a data-first program posture, and a structural mismatch in which lawfully-achieved Reduced Testing Allowance status makes many label failures non-actionable. The record does not adjudicate between them; carry all three or none. - **Figures and their bases:** $15,000 (the recurring stipulation fine level, and the speakers' shorthand; Bona Fides is exactly $15,000), $20,000 + surrender-as-revocation (GT Infusions stipulation), $23,750 (CC Brands stipulation), $200,000 + license loss (RM3 Labs, 2025; trade-press characterization of a listed 2025 action), $1,082,000 (MED 2025 annual report, total industry fines), half of 135 actions since 2023 (ProPublica; not independently re-counted), half of recent actions tied to adulteration (industry review presented at the February 2026 forum; unverified stakeholder figure). The third-party collection cost range (a modest bump to a tripling; the revised $40,000-$50,000/year for a daily-pickup operation) and the billion-dollar revenue loss are competing STAKEHOLDER estimates argued at the forum, never findings; the March summary carries the full spread and its fences. Keep each figure with its basis; never average or merge them. - **The chemicals box separates three claim types.** The identifications, classes, target pests, advisory counts, and prohibited status are from this site's desk reference built on the testing table (Rule 4-215(J)(8)(a)), the agriculture department's rule (8 CCR 1203-2 Part 17), and the 2026 advisory PDFs. The why-reached-for reading (rescue chemistry for mites and mold) is generic analysis of what these chemicals do; it is NEVER a claim that any advisory-named licensee acted deliberately, and the box says so in text. The prevention paragraph is a synthesis of the allowed list's character and standard cultivation practice, offered as context, not cultivation or compliance advice. "Most growers appear to be playing by the rules" is a hedged impression drawn from the record's scale (twenty contamination advisories and a handful of adulteration actions in eight months, against a market of thousands of licenses); "appear" must survive restatement, and the sentence is never a compliance statistic. The no-remediation-for-pesticides claim is Rule 4-240(B.5) verbatim (retest of the original test batch, then mandatory destruction) plus the rule's post-decontamination retest table, which includes no pesticide pathway. The dice-roll reading of the retest provision (worth using only near the threshold, where scatter can flip the verdict) is labeled analysis in text; using the provision is lawful, the provision's own purpose (protection against a single bad measurement) is stated alongside, and no data exists on how often it is invoked, so no usage claim is made. The fingerprint test (protection = challenges across all failure margins; dice roll = challenges clustered near threshold and disproportionately passing) is a stated prediction, not a finding; the discriminating data is not public as of publication and the page routes the question to the September 4 meeting. - **Staleness triggers:** any new MED final action citing rules 4-125, 3-1010, or 3-1015 or the adulteration statute 44-10-701(3); any potency/homogeneity health-and-safety advisory; any minimum-penalty rulemaking draft; publication of the CDPHE surveillance reports. Any of these should prompt an update of this page, including the credit if the Division delivers. ## Sources ### MED Final Administrative Actions (page and stipulation PDFs) - Role: primary; the controlling enforcement documents - URL: https://med.colorado.gov/final-administrative-actions - Used for: all 2026 action counts; CC Brands, GT Infusions, and Bona Fides case facts, charges, quotes, and sanction amounts; complaint numbers and their vintages - Factuality: authoritative for charges, admitted facts, and sanctions in settled cases; silent on cases still in progress, which the Division does not publish until final. PDFs are Google-Drive-hosted from the page; CC Brands and GT Infusions re-downloaded and quote-checked 2026-08-15. - Bias: not applicable; the documents are the object under discussion ### MED Health and Safety Advisories - Role: primary - URL: https://med.colorado.gov/health-and-safety-advisories - Used for: the twenty 2026 advisories and the stated cause of each; the zero-potency-advisories finding - Factuality: authoritative for what advisories were issued and why; an absence here does not prove absence of non-public remediation - Bias: not applicable ### MED Industry Bulletins (including 25-11 and 26-01) - Role: primary - URL: https://med.colorado.gov/industry-bulletins - Used for: the surveillance program's committed response ladder and reporting commitments (25-11); the Division's RTA-gaming candor (26-01); the absence of any 2026 bulletin on surveillance findings - Factuality: authoritative for what the Division committed to and announced - Bias: not applicable ### MED 2025 Regulatory and Enforcement Activity report - Role: primary - URL: https://med.colorado.gov/data-and-resources - Used for: the $1,082,000 total-fines figure and 2025 enforcement volumes - Factuality: authoritative for the Division's own 2025 totals - Bias: a self-report of the agency's activity; totals taken at face value, framing not relied on ### Colorado pesticide rules: 1 CCR 212-3 Rule 4-215(J)(8)(a) and 8 CCR 1203-2 Part 17 - Role: primary; the testing action-limit table and the agriculture department's pesticide-use rule for cannabis - Used for: the action-levels-are-analytical-thresholds characterization and the health-effects-unknown paraphrase (Part 17.02 states no pesticide has a cannabis tolerance and CDA does not recommend any untested product because "the health effects on consumers are unknown"); the chemicals box (prohibited status of all four detected compounds, the allowed list's character, and Rule 4-240(B.5)'s retest-then-mandatory- destruction fork for pesticide failures) - Factuality: authoritative rule text, staff-read in full with the adopted redlines diffed; the threshold-not-tolerance framing is this site's synthesis of the two rules read together - Bias: not applicable ### This site's forum transcripts and per-meeting summaries - Role: primary-adjacent; machine transcripts of the Division's published recordings, which are the source of record - Used for: the Greenman, Singer, and Division-response quotes with timestamps; the February adulteration-share figure; the March and June enforcement-alternative asks; the June edibles label-match figure - Factuality: transcription is machine-generated and can mis-hear; timestamps allow verification against the recordings - Bias: our own work product; the summaries carry their own appendix guards ### ProPublica, "Colorado Marijuana Regulators Consider Changes in Testing Sample Products" (2026-02-11) - Role: secondary; contemporaneous investigative coverage - URL: https://www.propublica.org/article/colorado-regulators-marijuana-lab-testing-system - Used for: the deputy director's "common violation" quote; the half-of-135-actions-since-2023 figure; Singer's "slap on the wrist" quote in print - Factuality: reliable for named quotes; the 135-action count is the outlet's tally and we have not re-counted it - Bias: investigative outlet with an accountability frame; no financial interest in the outcome identified ### MJBizDaily, RM3 Labs coverage (2025) - Role: secondary - URL: https://mjbizdaily.com/colorado-cannabis-testing-lab-rm3-potency-certification-suspended/ - Used for: the characterization of RM3's conduct as potency inflation and the $200,000/license-loss sanction - Factuality: headline-level access only (the site blocks automated full-text fetches); the action itself appears in MED's 2025 list, which is primary for its existence - Bias: trade outlet serving the industry; no specific interest in this lab identified