United StatesHirundo edits Chinese AI weights to remove censorship, it says.
Chinese open models trail US closed systems by 3% on benchmarks and lead Hugging Face downloads.
Hirundo’s chief executive Ben Luria says it edits open weights to remove state-aligned bias and dangerous knowledge, work he calls neurosurgery.
“We are bringing it up to speed on safety and to the standard of mainstream Western AI,” said Luria.
Europe’s OpenEuroLLM sovereign model is under development on Alibaba’s Qwen.
How each outlet framed it
- Rest of World
- centers machine unlearning as a commercial de-risking service, quoting Hirundo's CEO on bringing Chinese open-weight models to Western safety standards
Sources: Rest of World