Four AI Giants Just Put $10 Million Toward Keeping AI Safe

The Frontier Model Forum, a group formed by the world's biggest AI companies, announced a new AI Safety Fund. Google, Anthropic, Microsoft and OpenAI committed more than $10 million combined. The money will pay for independent research — meaning scientists who don't work for these companies — to study how to make AI systems safer. Think of it like four rival carmakers pooling money to fund traffic safety studies.

Why should you care? AI is already in your phone, your email, your search results and increasingly your doctor's office and workplace. Safety research answers questions like: can AI be tricked into giving dangerous advice? Can it be used to spread convincing fake news? Can it help hackers write attacks? These aren't science fiction questions — they're practical problems that show up as AI gets more powerful.

The fund is intended to pay outside researchers, usually academics and nonprofits, to find answers before problems happen. That's different from companies testing their own products, because outside scientists can publish findings freely, even when the results are uncomfortable for the funders.

The catch: $10 million sounds big, but it's a rounding error for companies spending billions on AI development — roughly what they earn in a few hours. And the companies paying for the research are the same ones whose products are being studied, which critics call self-regulation. Still, it's the first time fierce rivals have pooled money specifically for safety, and it sets a precedent other industries may follow.