Here is how it go:

• most expensive AI/ML engineer was layoffs.
• the spam filter was make by middle or even intern with vibe code
• Reddit want to focus on clean, non-bot, human content (to sell for AI training), so the go hard on bot
• clean data above all else, some mistake is acceptable.
• and again, maybe the intern who work on this see this model "good enough"
• leadership don't want to spend more money to prevent incorrect catch. People are going to appeal anyway, why have to pay more ?