senficon, 7 months ago

AI is giving #OpenSource a bad name, with the tired old arguments of security through obscurity. Just because open source alone does not guarantee the responsible development and deployment of an AI model, that doesn’t make it any less essential. The best guardrails will ultimately not stop a proprietary model from regurgitating the problematic inputs it was trained on, and intransparency over this process only seeks to obscure the problem. Open letter: https://open.mozilla.org/letter/