Big Tech AI launches are systematically destroying small language technology startups worldwide, according to AI ethics researchers Timnit Gebru and Abeba Birhane. When Meta released its No Language Left Behind translation model covering 55 African languages, investors told African NLP startups serving underserved communities to shut down immediately.
"[Investors] were like, 'Facebook has solved it, so your little puny startup is not going to be able to do anything,'" said Gebru, describing a pattern replicated across regions where minority language tech companies operate.
OpenAI representatives allegedly employ more direct tactics. "When they speak to people at OpenAI, they basically threaten them by saying, 'OpenAI is going to put you out of business soon because we're going to make our models better in your language,'" Gebru reported in a new AI Now Institute publication. The companies then offer to purchase these organizations' language data for minimal payments.
Birhane argues that 'AI for good' messaging serves as strategic deflection. "AI for good allows companies to say 'Look, we're doing something good! Everything about AI is not bad. And you can't criticize us,'" she said.
The researchers characterize dominant AI development as fundamentally extractive across global markets. "People came along and decided that they want to build a machine god. They end up stealing data, killing the environment, exploiting labor in that process," Gebru said.
Resource-intensive large language models concentrate power with well-funded corporations while making AI development inaccessible to organizations serving marginalized linguistic communities across Africa, Asia, and indigenous regions. The critique gains relevance as AI companies face regulatory pressure in multiple jurisdictions, with recent challenges from Anthropic on security classifications and UK trials restricting youth social media access.
The AI Now Institute publications represent growing international academic pushback against corporate narratives positioning AI development as universally beneficial while monopoly practices eliminate specialized alternatives serving non-English speaking populations.