Despite rapid generation of functional code, LLMs are introducing critical, compounding security flaws, posing serious risks for developers.
New translation models, open speech datasets, and automatic speech recognition benchmarks aim to expand AI support for African languages.