J Studios
The White House is preparing an executive order that would target what it deems to be “woke” artificial intelligence models, The Wall Street Journal reported, citing people familiar with the matter.
The order would require AI models of companies with federal contracts to be politically neutral, amid concerns that AI models may have liberal bias.
AI models, trained using information from the internet and other published sources, sometimes generate answers that are factually incorrect or misleading – a phenomenon called “AI hallucination.”
For example, Google (NASDAQ:GOOG) (NASDAQ:GOOGL) came under fire last year as its AI assistant Gemini showed a black George Washington and racially diverse Nazis.
This trend has troubled White House officials, including AI Czar David Sacks – one of the architects of the executive order targeting “woke” AI. But the order may be controversial, given concerns in the tech industry that the government could play favorites.
xAI – owned by Trump’s on-again, off-again ally Elon Musk – prioritizes being “anti-woke,” but recently faced criticism after its Grok chatbot generated antisemitic content.
Dario Amodei, CEO of Amazon (NASDAQ:AMZN)-backed Anthropic, recently criticized a Republican proposal to bar states from regulating AI for 10 years. Anthropic has also hired former Biden administration staffers.
To note, OpenAI, Google (NASDAQ:GOOG), Anthropic, and xAI this week secured contracts worth up to $200M from the Pentagon to expand the federal use of their AI tools.
The Trump administration is expected to unveil several executive orders next week, including the one on “woke” AI, aimed at ensuring the U.S. maintains its AI edge over China.
The orders would aim to promote exports of American chips and AI tools, and accelerate permitting for data centers, sources told WSJ.