The Guardian on Respond’s language rights interventions: “Lost in AI translation”
Respond Crisis Translation shared with The Guardian about how the U.S. government’s growing reliance on unsupervised machine translation to cut costs has jeopardized several asylum applications.
We’ve successfully intervened in several egregious cases:
Our client Carlos, an afro-Indigenous asylum seeker from Brazil who speaks Portuguese, sought our support after having been deprived of language access for months in U.S. custody without proper medical care.
A machine mistranslated an Afghan woman’s asylum affidavit, changing all cases of “I” to “we.” Her application was rejected.
An asylum seeking domestic violence survivor described abuse by “mi jefe,” a common colloquialism to refer to one’s father. She was initially denied asylum after AI translated the term to “my boss.”
These cases were only won because Respond’s well trained, trauma-informed human translators intervened.
“It’s almost impossible for a machine to convey the same message that a professional interpreter with awareness about the country of origin can do, including cultural context,” Respond’s Afghan Languages Team Lead Uma Mirkhail says.
In the past month alone, over 180 Afghans won their asylum claims after Uma and her team mobilized around the clock to translate and interpret legal clinics. Among the successful cases was an asylum seeker whose application had previously been denied due to the government’s machine translation errors.
“[The] government will frequently weaponize small language technicalities to justify deporting someone. … The application needs to be absolutely perfect,” Respond’s founder Ariel Koren said, emphasizing that the government weaponizes its very own translation errors to impede access to asylum.
Read more about the fight we are up against in The Guardian.
Donate
Press inquiries: media@respondcrisistranslation.org
