CEO Dario Amodei argues the tech industry shouldn't assume that making open AI models will benefit the good guys and the bad guys equally. In fact, it might skew more toward the latter.

Dozens of AI companies, led by Nvidia, said on Monday they had formed a coalition called the Open Secure AI Alliance, to develop open-source AI tools for defensive cybersecurity. Days earlier, on Friday, many of those same companies signed an open letter urging the U.S. government not to ban open-weights AI models.

A16z, Dell, Microsoft, Meta, Nvidia and Palantir are among over 20 companies that signed onto the letter shared exclusively with POLITICO.

Kari Briski, Nvidia’s vice president of generative AI software, sees open AI models as a way for companies to iterate their way to AI success by adapting the models with their own data.

As Chinese AI models grow in capability and popularity among US companies, the arguing over what should be done about them has reached a fever pitch.

OpenAI, Anthropic, and other Western labs have spent billions training the model weights at the heart of their most advanced systems.

Defining and evaluating openness in AI requires examining both the model and system stack, including interfaces, safeguards, infrastructure, and deployment practices

Unlike Meta's previous Llama models, the latest Muse Spark AI models are closed. That decision was made to help the Meta Superintelligence Labs team speed up model development and skip the extra engineering required to release a model publicly.

For companies that want control over data, model behavior and fine-tuning, that smaller footprint may be more important than chasing the highest possible benchmark score.

The future of AI may be dictated not by who builds the smartest model, but who builds the models everyone runs on. Right now, that's China.

A ban on open Chinese AI models under the guise of security might actually put our cybersecurity more at risk.

Closed systems create friction that hinders adoption and customization. Kimi K3's recent release suggests that the market may ultimately favor open architectures.