Florida's New Breath Test Machine Is Here. It Solves Some Problems — and Creates New Ones.

If you're arrested for DUI in Florida this year, there's a real chance the machine measuring your breath alcohol isn't the one your lawyer has spent years learning to cross-examine. The Florida Department of Law Enforcement is in the middle of a statewide swap, retiring the Intoxilyzer 8000 — in service since 2006 — in favor of its successor, the Intoxilyzer 9000, built by the same manufacturer, CMI, Inc. Agencies across the state, including several in the Tampa Bay area, have already taken delivery of the new units and are running officers through certification courses. FDLE expects it to become the primary evidentiary breath-test instrument statewide sometime this year.

The state will market this as a straightforward upgrade: newer hardware, a touchscreen instead of a fluorescent two-line display, better data transmission. Some of that is true. But a device swap of this size — new hardware, a new training curriculum, a new certification course, agencies running old and new machines side by side during the transition — is exactly the kind of moment that creates openings for the defense, not just the state.

What the 9000 Actually Improves

To be fair to the new machine, several changes are genuine upgrades over a seventeen-year-old design:

• A modern interface. The 8000 ran on a two-line, twenty-character fluorescent display and a physical keypad. The 9000 uses a color touchscreen with a barcode scanner and magnetic-stripe card reader, which cuts down on the kind of manual data-entry mistakes — transposed numbers, wrong case IDs — that used to give defense attorneys an easy opening.

• Better connectivity. The 8000 reported results over an internal analog phone modem. The 9000 uses Ethernet and USB with configurable network settings, which should mean faster, more reliable transmission of results to FDLE and fewer lost or corrupted data files.

• More optical detectors. The 9000 reportedly uses four pyroelectric infrared detectors, each paired with its own filter, versus two on the 8000 — in theory, a finer-grained spectral read on the breath sample.

• Improved RFI shielding. Radio frequency interference from police radios and cell phones was a known vulnerability on the 8000; the 9000's manual describes improved shielding, even though the interference error code hasn't gone away entirely.

None of that is nothing. But several of the changes that make the 9000 cheaper and faster to run also strip out safeguards that used to give both sides — and juries — more insight into whether a given result was trustworthy.

What the 9000 Gets Worse — and Why It Matters for Your Case

It calibrates against gas, not liquid. The 8000 required monthly inspections using "wet bath" simulators — liquid ethanol solutions that vaporize the way alcohol actually does in a human lung, governed by Henry's Law. The 9000 has scrapped wet-bath calibration for routine inspections entirely, relying only on pre-mixed dry gas from a canister. That's cheaper and easier to standardize, but it also means the monthly check no longer tests whether the machine can accurately read alcohol vapor coming off a liquid source the way a real breath sample does — which is precisely the physiological process the machine is supposed to be measuring.

Its certified range got narrower, not wider. The 8000 reported quantitative readings up to 0.600 g/210L before flagging an out-of-range error. The 9000's certified range tops out at 0.300 g/210L. Any result above that is outside the instrument's verified accuracy window — which matters a great deal in the higher-BAC cases that carry Florida's enhanced penalties.

It tells you less when something goes wrong. The 8000 distinguished between a falling reading (SLOPE NOT MET) and a rising, unpeaked one (SLOPE NOT LEVEL) — a distinction that mattered because a rising, un-peaked sample is a classic signature of mouth alcohol contamination rather than a true lung-air reading. The 9000 collapses both into a single generic "SLOPE NOT MET" code, which means less diagnostic information reaches the printout — and less for either side to work with in figuring out why a sample was rejected.

Operators are trained to know less about the hardware, not more. The 8000 curriculum taught agency inspectors basic hardware diagnostics — fuses, printer mechanisms, physical components. The 9000 curriculum explicitly bars inspectors from opening or servicing the unit at all. That may reduce the risk of an untrained officer breaking something, but it also means the person certifying the machine as reliable, and potentially testifying to that in court, is trained to be an operator of a sealed box rather than someone with any working understanding of what's happening inside it.

Mouth alcohol, chemical interference, and RFI are all still live issues. Acid reflux or a recent burp can push residual stomach alcohol into the mouth and spike a reading; the histogram is supposed to catch it, but isn't foolproof. Hand sanitizer or bleach-based cleaning products used near the machine can trigger an ambient-air failure. High-output radios or nearby cellular devices can still trigger an RFI error. None of these are new to the 9000, but they remain fully available defenses regardless of which machine was used.

For anyone facing a DUI charge during this transition period, there's an additional, very practical angle: courses and certifications for the two machines are not interchangeable, and agencies are running mixed fleets while officers work through new training requirements. That makes it worth confirming, case by case, that the specific machine, the specific operator's certification, and the specific agency inspector's permit all match up correctly for the device actually used.

Florida Has Been Here Before — and Just Did It Again

This isn't the state's first fight over what's inside a CMI breath-test machine, and it isn't even the most recent one.

Back when Florida relied on the Intoxilyzer 5000, roughly 150 Florida defendants petitioned the courts to force disclosure of its source code, pointing to CMI's history of making changes to its machines without proper certification and at least one past recall tied to faulty software.

Then, this August, a three-judge panel in Duval County went straight at the 8000's credibility. The panel found that the Jacksonville Sheriff's Office and FDLE's own Alcohol Testing Program had been replacing worn O-rings on Intoxilyzer 8000 units and logging the work as routine maintenance — when Florida's rules only allow the manufacturer or an authorized repair facility to perform an actual repair. Because swapping the O-rings was a repair, not maintenance, the state couldn't establish the "substantial compliance" Florida law requires before a breath result is admissible, and the panel suppressed the test. The ruling extended a 2025 Second District Court of Appeal decision, State v. Kilburn, which threw out 8000 results on the same theory after an FDLE inspector replaced a broken breath tube instead of sending the machine to an authorized repair facility. The Duval ruling alone put roughly 950 pending DUI cases in Northeast Florida under review.

What that means for the 9000. The O-ring problem was possible because 8000 inspectors were trained to do exactly that kind of hands-on hardware work — fuses, physical components, and, evidently in practice, worn internal parts. The 9000 curriculum closes that particular door: inspectors are explicitly barred from opening or servicing the unit at all. In theory, that should make an O-ring-style violation harder to repeat, since there's no field-level repair left for an inspector to mislabel as maintenance — a malfunctioning 9000 arguably has to go back to CMI or an authorized facility, which is the outcome Florida law was already supposed to require.

That's worth treating with some skepticism as a fix, though. The same sealed design that removes one failure mode reintroduces the one flagged above: inspectors who can't open the box also can't meaningfully evaluate what's actually wrong with it, which pushes reliability questions further out of anyone's reach rather than resolving them. A sealed unit doesn't stop an agency from being tempted to keep a failing machine in service — it just changes what a defense attorney needs to request in discovery to catch it. If the Kilburn/O-ring pattern shows up again on the 9000, it's more likely to surface as a repair paper trail that doesn't match the machine's actual service history, or as a unit that should have hit its automatic "Disabled Mode" lockout but didn't, than as a literal parts swap logged as maintenance.

How Other States Have Fought Breath-Test Machines — and What Actually Worked

Florida isn't unique in confronting a "trust the black box" instrument, and looking at how other states have handled it is instructive, because the strategies that work depend heavily on each state's rules of evidence.

New Jersey forced the code into the open — and it wasn't pretty. After a two-year fight, defense attorneys in State v. Chun got the New Jersey Supreme Court to order Draeger, maker of the Alcotest 7110 used statewide, to turn over its source code. When independent analysts reviewed it, the code was described as patchwork, built up over many revisions with no consistent coding standard — and notably, that assessment came not just from the defense's expert but from the manufacturer's own reviewing lab as well. It's one of the clearest examples of a source-code fight actually validating what defense attorneys had argued for years: that "proprietary" and "reliable" aren't the same thing.

Minnesota went after the same manufacturer Florida uses. In State v. Underdahl, the Minnesota Supreme Court ordered CMI to produce the source code for the Intoxilyzer 5000 — the same company that builds Florida's 8000 and 9000. It's a useful precedent specifically for Florida cases, since it establishes that courts have compelled this exact manufacturer to open its code before.

Washington shows the fight isn't over even when you win access. When Washington state adopted Draeger's Alcotest 9510, independent researchers who got court-approved access to the source code found flaws they said could produce inflated readings. But their final report was never finished — Draeger accused one researcher of violating the protective order governing how the code could be used and threatened legal action, and the analysis stalled. It's a reminder that winning the right to see the code doesn't guarantee a clean, usable result; manufacturers still have tools to slow the process down.

North Carolina shows why this strategy doesn't travel everywhere. North Carolina uses a different machine, the Intox EC/IR II, but more importantly, it has a state statute making breath results automatically admissible as long as testing follows Department of Health and Human Services rules. That kind of statute blunts the practical value of a source-code challenge, because even a demonstrably imperfect result can come in without the same evidentiary fight defendants get in states like New Jersey or Minnesota.

The lesson for Florida cases isn't that source code litigation is automatically the right move here — Florida's evidentiary framework is its own, and that kind of fight is a significant undertaking. It's that the underlying skepticism about black-box breath-testing devices has a track record of being vindicated, and CMI specifically has been on the losing end of it before.

What This Means If You're Facing a DUI Charge Right Now

A number on a printout can look final. It isn't. Whether the machine was an 8000 or a brand-new 9000, whether the operator's certification matches the device actually used, whether the monthly inspection reflects a real calibration standard or just a canister of gas, whether the "SLOPE NOT MET" flag on your test masks something the older machine would have identified more specifically — these are all questions with real answers, and Florida's transition year is an unusually good time to ask them. If you've been arrested for DUI in Florida, talk to a criminal defense attorney who understands the equipment behind the number before assuming the reading is the end of the story.