Most combustion will produce some noxious byproducts, and absent the very expensive catalyst mentioned in TFA, ammonia also releases high levels of NOx (nitrous oxides).
CO has a lower possibly lethal concentration of about 200 ppm (with long exposure), though higher levels will cause more rapid and severe issues, with levels from 800 to 12,800 resulting in death in from 2--3 hours to a few minutes. Auto exhaust concentrations without a catalytic converter are 35,000 ppm.
https://www.abe.iastate.edu/extension-and-outreach/carbon-mo...
In the case of ammonia, it is the fuel itself which is toxic.
LC50 (lethal concentration, 50% mortality) varies by duration of exposure, from 40,300 ppm (10 min) to 2,000 ppm (4 hr). LC50 for 30 minutes exposure ranges from 200--2,000 ppm.
https://www.cdc.gov/niosh/idlh/7664417.html
Note that the kill mechanisms differ markedly. CO binds to hemoglobin, preventing oxygen from doing so, and effectively asphyxiating the victim. Treatment with oxygen should promote recovery in most cases. Ammonia is chemically caustic, exothermic when hydrating, and chemically induces necrosis and tissue death, most notably in the lungs if inhaled. Treatment and recovery are far more critical than with CO.
https://www.ncbi.nlm.nih.gov/books/NBK546677/?report=classic
On top of that, CO is a fractional exhaust product whilst ammonia is the hyperabundant fuel. Any fuel spill or partial combustion presents and immediate and severe inhalation threat.
It's best to avoid both. Ammonia strikes me as the markedly greater risk.
With exposure to organic combustion virtually universal, CO poisoning deaths in the US range about 438/yr (1999--2012), or 1.48/million. From the accompanying map of case intensity, I'd surmise indoor heating is a principle risk case, with cases concentrated in northern and Rocky Mountain states. On balance, that's a fairly low per-person risk, the overall incidence rate is a measure of net exposure.
https://pubmed.ncbi.nlm.nih.gov/26032660/
In my earlier comment, CO has nearly exactly twice the incidents as ammonia, though I strongly suspect that this is based on a vastly larger potentially-expose population. Even with very limited exposure to potential poisoning, ammonia represents an absolute risk half that of CO, which has universal exposure.