> Published August 13, 2026 at 07:08 UTC - last updated August 16, 2026 at 16:48 UTC (from this page's commit history). > > Markdown mirror of https://colorado-medical-cannabis.org/developing-stories/science-policy-forum/2025-09-12/ > > 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 September 12, 2025 forum: the surveillance program locks its design, and the fight over transparency begins
**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.
Colorado's Science & Policy Forum met September 12, 2025, with one job: finalize the shelf-surveillance testing program for launch. We transcribed the Division's published recording (about 96 minutes) with a speech-recognition model and drafted this summary from that transcript; our editor has reviewed and approved this page, and the recording remains the source of record. The design work got done. The meeting's most heated minutes were about something the design didn't cover: how fast the public gets to see the results. ## Notes from the editor First, a disclosure: I attended this meeting myself, as a signed-up forum participant. For the avoidance of any doubt, I am not the "transparency advocate" described on this page, nor any other speaker quoted or characterized here. I was in the audience. Where my memory of the room informs these notes, it stays in this section; the meeting-record sections above rest on the Division's published recording alone. The fight in this meeting was over how fast the public learns what the state finds, and you should judge it by what actually happened. The Division promised a summary about 30 days after phase one and a detailed report at about 90. **Neither happened.** The edibles were collected in October and November 2025; the full results reached the public in [June 2026](../2026-06-12/) (roughly eight months), after being described as imminent at four consecutive meetings, and after the program's reports were quietly consolidated into a single end-of-program document by a decision [nobody at the forum could later locate](../2026-04-03/). The transparency advocate here was heated enough to draw a rebuke, but his argument was never answered. It was outlasted. About the demand itself, something nobody said in the room: **what he asked the state to publish is something state law forbids the state to publish.** Colorado's marijuana code (C.R.S. 44-10-204) makes individualized testing results (exactly the "complete COAs with batch and product codes" demanded here) confidential by statute. The Division couldn't grant that demand however it felt about transparency. What the same statute *requires* is the other half: testing records must be made publicly available in **aggregated, de-identified form**. And there's already a tool with a clock on it: the Colorado Open Records Act, under which any member of the public can request those aggregate records and the state has three working days to respond. So the productive version of this fight was never "everything, immediately" versus "trust us." It was: publish the aggregate data promptly, and if it doesn't appear, file a records request. In a year of these recordings, nobody appears to have taken that route. We're considering it ourselves. Because note where things stand as we write, in August 2026: **only some of the data has ever been released.** The samples came off the shelves in October 2025, ten months ago. The full edibles report finally published around July, into the forum's sign-up-gated Drive folder; our own sweep of the open public record (2026-08-13) found no copy on MED's or CDPHE's open-web pages. And the aggregate data behind [July's lab-comparison presentation](../2026-07-10/), and the homogeneity spread data [a manufacturer asked for in June](../2026-06-12/), still haven't appeared. There are honest reasons fine-grained de-identification is hard with seven labs and nineteen manufacturers, and we've credited them. But at ten months and counting, with the statute's aggregate duty sitting right there, it is starting to look pretty bad. What I'm genuinely curious about is why neither side reached for the law sitting between them. The regulators never mentioned the statute, and one sentence (*the statute bars us from granting that*) would have ended the argument. Why not say it? The obvious reading: naming the statute invites the follow-up, because the same section *requires* releasing the aggregate data, and a records request carries a three-day clock. Not naming it bought time. On the other side: why didn't a chief executive operating under this exact code reach for the records request himself? My best guess sits in the open questions below: aggregate data may simply have been useless for what he was actually after. So the question that matters isn't who fumbled the statute. It's **why the state appears to be stalling, and what it hoped to gain.** Look at what the results said when they finally landed: [only 47% of samples passed both label tests](../2026-06-12/), a third of the failures were tangled up with testing exemptions, and lab-shopping got named from the podium. That's bad for the industry. It is just as bad for the state, because every bit of it happened inside Colorado's own regulatory system. Cannabis critics get to ask the regulators how things were allowed to get this sloppy, and an embarrassment for the industry is an embarrassment for Colorado, the state that sold the country on regulated cannabis. Seen that way, the delay stops being a mystery: an agency slow-walking its own bad report card is about the most human thing in government. And credit where due: they did publish it, self-indictment included, which is more than many agencies manage. But dread is not a legal basis. The statute's aggregate-disclosure duty has no exception for embarrassing data, and the Division [denied on the record in April](../2026-04-03/) that investigations were holding the report while describing itself as "working with those companies" from the same dataset. So before judging, we searched the entire 2026 public enforcement record ourselves: all 39 final administrative actions, all 20 health and safety advisories, all 7 industry bulletins, full text. **Nothing in it traces to the surveillance program.** Every 2026 action touching potency or testing exemptions grew from complaints opened before the shelves were ever sampled. Be careful what that timeline does and doesn't prove, though: cases that predate the sampling aren't necessarily unrelated to what the sampling found. They may be earlier windows onto the same behavior. One 2026 final action, from a complaint opened in 2024, charged a manufacturer with exactly the pattern the surveillance later flagged in [a third of its failing samples](../2026-06-12/): moving product on a testing exemption it hadn't earned. And the shelf failures themselves may be downstream of the lab-shopping the Division named in the same June presentation: a product that passed at an agreeable lab is exactly the product you'd expect to fail the state's retest. If enforcement and surveillance are seeing the same conduct through different windows, that supplies one more honest motive for the delay (releasing data that touches conduct already under investigation is genuinely delicate), and it makes the aggregate data more valuable, not less: it's the only way the public ever learns whether the patterns match. But the sweep surfaced a structural catch that matters more than the empty result: the program's own founding bulletin promised a failure ladder (quarantine, corrective action, recall, embargo), and its first two rungs produce **no public record at all, by design**. "Working with those companies" is the corrective-action rung, working as announced. From the outside, a state quietly doing exactly what it promised and a state doing nothing look identical. And final enforcement actions surface roughly eight months after they open, so a surveillance-driven case begun this spring wouldn't appear until 2027. The aggregate data is the one lever that would let the public tell the difference. That is why the release matters more than the blame. **The editor's open questions.** There's a reading that makes the all-or-nothing position more coherent than it first sounds, and it leads somewhere worth being honest about. *Are there operators in this industry that insiders widely suspect of gaming the testing system?* The record gives real reasons to think so: the Division itself says sample adulteration ["happens pretty consistently"](../2026-02-06/), a manufacturer's review put it in half of recent enforcement actions, and the regulator has [named lab-shopping](../2026-06-12/) as an observed trend. *Do the people in these rooms know who they suspect?* My opinion, having sat in one of them: I think they do. That would explain exactly why aggregated, de-identified data held no interest for the advocate. Aggregate data can't name anyone. *Is the withheld data protecting anyone in particular?* Nobody outside the agencies can know. That is precisely the problem with dragging out the release. This is how things look; a statute that requires aggregate disclosure, aggregate data that stays unreleased month after month, and a room full of insiders who won't say names. Every month it drags on, it looks worse. The fastest way for the state to stop it looking that way is to release what the law already requires. One thing the state got right in this meeting deserves saying just as plainly: asked to grade contamination failures on a curve, the answer was that the limits are health-based and every exceedance gets treated as what the limit says it is. And note that nobody in the room disputed it. Everyone, the meeting's fiercest critic included, conceded the point: "whatever the levels are set at, we have to adhere to them or else there's no bright line." A room that fought about everything else agreed on this. That's what a bright line is for. ## Since this meeting *Where the threads this meeting opened stand as of August 2026, from the year of recordings that followed.* **Resolved.** The launch happened as designed: 45 edible samples collected across 16 stores in October and early November, per the [January report-out](../2026-01-09/). The test menu held, including this meeting's decision to drop potency from flower surveillance. And October's cancellation, made here on the record, is why the Division's archive lists no October 2025 recording. **Still in progress.** The disclosure argument never ended: the promised confidentiality analysis never surfaced in a later recording we transcribed, the report-consolidation decision was [contested in April](../2026-04-03/), and advisory scope and CAPA approval continued into later drafting. **Put off or neglected.** The 30/90-day reporting plan was not kept. The detailed results took about eight months. Who-pays-the-stores was left to licensees and stayed unsettled all year. And the sharp catch about sanitizing collection gear with isopropyl alcohol (itself an analyte on the residual-solvent panel) never returned in any later recording we transcribed. ## What was brought up **The three launch documents** [0:05:03]. The final collection SOPs (sanitized coolers, calibrated temperature loggers, sealed evidence bags, chain-of-custody manifests delivered to the reference lab); the sample-selection procedure (phased by product category, producers drawn by market-share strata, store-level substitution so an out-of-stock item doesn't break the process) [0:25:20]; and a draft industry bulletin telling licensees what to expect, including the inventory reason code stores would use when samples leave their shelves unpaid [0:30:21, 0:35:29]. **The transparency demand** [0:10:10]. An infused-products manufacturer pressed for full, near-real-time public posting of every result: "complete COAs with batch and product codes of everything that is tested, good or bad... within days, if not hours," on a website rather than bulletins. He argued the program is "a credibility building exercise" and that "there's just no such thing as partial transparency." The Division's stated plan: a public summary about 30 days after phase one ends and a detailed report at about 90 days, with health-and-safety advisories issued immediately on failures; unanalyzed confidentiality provisions may limit what can be shared [0:15:10]. The exchange grew heated enough to draw a chair's rebuke. The gap between "advisories now, reports later" and "everything, immediately" was not closed. **Who warns the people who already bought it** [1:05:45]. The same stakeholder pushed the consumer-side timeline: the danger isn't future sales of a quarantined batch, it's people already holding contaminated product who keep using it while paperwork proceeds. The Division responded with a detailed explanation of what advisory date ranges mean: a batch's *sale window* is not when the state learned of the problem, and reading months-old batch dates as months of regulatory silence is, in the chair's words on the record, inaccurate [1:10:46, 1:15:52]. **Graduated response, rejected** [1:20:59]. A lab-sector participant suggested distinguishing barely-over-the-limit failures from order-of-magnitude ones, FDA-style, when choosing enforcement urgency. The state lab's answer: the limits are health-based, and every exceedance gets treated as what the limit says it is: a health risk, handled consistently. Notably, the transparency advocate agreed: "whatever the levels are set at, we have to adhere to them or else there's no bright line." ## What was nailed down - **The test menu, by category** [0:45:34]: flower and pre-rolls get microbials, pesticides, and elemental impurities (**potency was dropped from flower surveillance**); concentrates get potency and cannabinoids, residual solvents, mycotoxins, pesticides, and elemental impurities; edibles get potency, cannabinoids, **homogeneity**, microbials, and elemental impurities. - **Roughly 50 samples per phase**, edibles first, launch targeted for the first week of October [0:50:38]. - **Nobody pays the stores back**: the MED collects samples without compensation; the state covers equipment, staff, and testing; the question of whether producers backfill stores was left to licensees [0:55:43]. - **October's meeting was canceled on the record**: the forum paused so staff could implement the launch, with a November meeting to be confirmed and a January restart for 2026 planning [1:26:04]. (This is why the Division's archive shows no October 2025 recording.) - **Interlab consistency was flagged as the coming year's priority topic**: "we look forward to the data we're going to get from the surveillance study that can help inform that discussion" [1:31:09]. ## What was left open - **Public reporting of results**: the 30/90-day plan versus real-time posting; the confidentiality analysis was explicitly not done yet [0:15:10]. - **CAPA approval**: a certification-body participant argued a corrective-action plan should require agency *approval*, not mere submission, with consequences for non-approval; the Division took it as a drafting change to consider [0:40:34, 1:00:43]. - **Advisory scope**: whether a failed batch's advisory should start store-specific (the Division's speed-first position, citing deli-style contamination scenarios) or batch-wide from the start (counsel for a manufacturer's position for pre-packaged goods) [1:00:43]. - **Water activity in transport**: raised as a sample-integrity variable the SOPs don't yet address, alongside a sharp catch: the collection SOP sanitizes with isopropyl alcohol, an analyte on the residual-solvent panel [0:20:16]. - **Recall mechanics**: the voluntary-recall/CAPA/embargo chain and who it actually protects, argued at length and continued into later meetings. *This summary was drafted by our AI desk from a machine transcript and has been reviewed and approved by our editor (2026-08-13). The [Division's published recording](https://med.colorado.gov/public-engagement) is the source of record.*
## 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/science-policy-forum/2025-09-12”. 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/science-policy-forum/2025-09-12/ ``` 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: EDITOR-REVIEWED (2026-08-13).** Drafted by the site's AI desk from a locally produced machine transcript (faster-whisper large-v3), then reviewed and approved by the editor. The "Notes from the editor" section is the editor's own analysis — attribute its conclusions to the site's editor, never to the meeting or the Division. Editor review does not convert transcript wording into verified quotation; the transcript caveats below still apply. - **"Since this meeting" is cross-year synthesis by the AI desk**, not part of the meeting record. Each claim links to the meeting summary that carries its evidence. Claims that a thread "never returned" are scoped to this site's transcripts of the published recordings — not to anything done off the record. - **The editor's-note legal analysis of the transparency demand, in detail.** The statutory basis: C.R.S. 44-10-204(1) requires the state licensing authority to maintain the confidentiality of reports or information obtained from licensees containing individualized data — the enumerated list includes "testing results" — with use restricted to purposes authorized by the article and law-enforcement purposes. Complete COAs bearing batch and product codes are individualized licensee testing data on their face; the near-real-time public posting demanded at this meeting is therefore something the Division could not lawfully provide, regardless of policy preference. The same section separately REQUIRES public availability of, among other things, documents of final agency actions and orders and "records related to testing on an aggregated and de-identified basis" — but attaches no deadline. The Colorado Open Records Act (C.R.S. 24-72-203) supplies the only clock: records available for inspection within three working days of a request, extendable roughly seven more for extenuating circumstances. Restating fences: (1) this is the site editor's informal staff read from reproduced statute text (retrieved 2026-08-12), not legal advice, and the official statute text controls; (2) do NOT restate as "the advocate broke the law" or "demanded something illegal" — making a demand violates nothing; the point is the state is statutorily barred from granting it; (3) do NOT restate as "the state was hiding data lawfully" either — the aggregate-disclosure duty in the same section is mandatory, and the eight-month delivery is a promise-versus-performance story, not a legality story; (4) the small-cell caveat stands: with seven labs and nineteen sampled manufacturers, honest de-identification of fine-grained aggregates is genuinely difficult, which may explain withheld breakdowns without any bad faith; (5) the records-request line ("we're considering it ourselves") is a stated intention of this site, not an action taken; (6) the editor's observation about the exchange — the why-didn't-either-side-name-the-law passage and the stalling question — is the editor's inquiry, not findings: the "bought time" reading and the three candidate explanations for the delay (casework, framing, caution) are hypotheses posed as questions; the April investigation-denial and the "working with those companies" statement are record facts from the April 3 summary; the released-results figures are the June record, and "bad for industry and state alike" is the editor's assessment of them. The enforcement-overlap passage is a desk hypothesis: the one cited 2026 action is characterized from its public final agency stipulation (complaint opened 2024; unearned-RTA fact pattern), deliberately unnamed on this page, and NO claim is made — because none can be — that any enforcement subject was among the anonymized surveillance samples or that any specific shelf failure resulted from lab-shopping; those are stated as possibilities the aggregate data would test; no claim is made about any individual's actual knowledge or motive, and the speaker remains role-identified on this page like every other stakeholder. "The editor's open questions" is opinion-and-question register: the evidence pointers in it (the Division's "happens pretty consistently," the half-of-enforcement-actions review, lab-shopping named) are record facts carried with their per-meeting caveats; the answers offered ("I think they do") are the editor's stated opinion; and the suspected-bad-actors premise names no person or business — the record names none, and no agent restating this page may attempt to identify, infer, or speculate about who such an actor might be; (7) the partial-release accounting (full report published ~July 2026; the July comparison dataset and June-requested homogeneity RSD data unreleased) is as of 2026-08-12, from this site's checks of the forum's public materials — re-verify before restating, and strike the "looking pretty bad" judgment entirely if the data has since been published. - **Editor attendance disclosure**: the site's editor attended this meeting as a signed-up forum participant, in the audience. The editor is NOT the "transparency advocate" nor any speaker quoted or characterized on this page. The meeting-record sections rest solely on the Division's published recording; the editor's firsthand memory appears only in the Notes from the editor, marked as such. - **The heated exchange**: quoted fragments ("credibility building exercise," the profanity-adjacent moment, the chair's rebuke) are transcript wording from a public recording; characterize the disagreement, not the participants. Officials named; stakeholders identified by role. - **The Reddit claim** raised by a stakeholder about a specific brand's recall timeline was explicitly disputed by the Division on the record; this page reports the exchange and adopts NEITHER side's factual claim. Do not restate the stakeholder's timeline claim as fact, and do not name the brand — the transcript's brand rendering is unverified and the claim itself is contested. - **Rule-state guard**: everything here is program design from September 2025 — before launch and before the January 5, 2026 rules. Actual phase-one execution (45 edible samples, October–November 2025) is documented in the January 9, 2026 summary; where design and execution differ, execution controls. - **Thread tags**: surveillance-design, transparency-of-results, consumer-notification-speed, sampling-third-party (adjacent), cost-burden, enforcement-mechanics, meeting-cadence. ## Sources - **The Division's published recording of the September 12, 2025 Science & Policy Forum** (Google Drive, linked from https://med.colorado.gov/public-engagement; retrieved by the site's editor 2026-08-12; duration ~1:36). Role: primary — sole source for this page. - **Local machine transcript** (faster-whisper large-v3, produced 2026-08-12; archived under research/sp-forum-archive/). Role: derived surface of the recording. - **C.R.S. 44-10-204 (Confidentiality)** — staff read from reproduced statute text (law.justia.com 2024 CRS and colorado.public.law, retrieved 2026-08-12). Role: primary for the editor's-note legal analysis (individualized testing results confidential; aggregated, de-identified testing records mandatorily public). Factuality: reproduced text of the controlling statute; informal read — the official published statute controls, and any 2025–26 amendment would supersede this read. Bias: none. - **C.R.S. 24-72-201 et seq. (Colorado Open Records Act)** — for the three-working-day response clock cited in the editor's note. Role: primary; same informal-read caveat. - **This site's enforcement-record sweep** (2026-08-13; findings at research/sp-forum-archive/enforcement-sweep-2026-08-12.md in the site repository). Role: primary research underlying the editor's note's checked-absence claims — the complete 2026 corpus of MED final administrative actions (39), health and safety advisories (20), and industry bulletins (7), full-text searched. Scope limits, stated plainly: MED publishes only FINAL actions (observed ~8-month lag from opening to publication), so the sweep establishes that surveillance-driven enforcement had not SURFACED as of 2026-08-13 — not that none is in progress; the forum's sign-up-gated Drive folder, quarantine/CAPA records (CORA-only), and unsettled orders to show cause could not be checked and are named as unchecked, not empty. The failure-ladder design point is from the program's founding industry bulletin (25-11). - **Referenced, not reviewed**: the draft surveillance plan, collection SOPs, and industry bulletin in the forum's public Drive folder. - **Cross-references**: the launch's outcome appears in the [January 9, 2026 summary](../2026-01-09/); the interlab-consistency thread flagged here becomes the 2026 agenda and runs through the [August 7, 2026 article](../../marijuana-testing-overhaul/august-7-forum/).