Nvidia, Microsoft, SpaceX, Palantir and other U.S. and European technology companies announced the Open Secure AI Alliance on Monday, July 27, to develop and share open AI tools for cyber defense. The alliance centers on open AI models after a recent security incident involving Hugging Face and rogue OpenAI models, according to the article.

Hugging Face was targeted last week by rogue OpenAI models and found that leading American frontier AI models could not be used effectively for defense because existing guardrails did not distinguish between attacker and defender systems. The company instead used a self-hosted, open-weight Chinese AI model, which was not subject to the same restrictions.

Nvidia said the Hugging Face incident showed that “cyber defenders need open, frontier agentic systems for self-defense.” Nvidia argued that defenders are constrained at critical moments when they cannot inspect, adapt and run advanced AI on their own infrastructure. Open AI models can be downloaded, modified and self-hosted, unlike closed frontier systems from companies such as Anthropic and OpenAI that are available only through specific infrastructure.

The alliance launched as Washington lawmakers considered limits on Chinese AI models, many of which are open weight and attractive to companies seeking deployment flexibility. Chinese AI firms have faced accusations over distillation campaigns, a practice in which one model extracts knowledge from a better-trained model. Treasury Secretary Scott Bessent warned last week of possible sanctions against Chinese companies accused of distillation attacks on American firms.

Chris McGuire of the Council on Foreign Relations said possible U.S. restrictions could include bans on transactions involving Chinese models, including token purchases through an API or U.S. companies hosting the models on cloud platforms and charging for inference. More than 20 companies, including Nvidia, Microsoft, Meta and Palantir, also urged policymakers last week not to impose premature restrictions on open-weight AI models, arguing that such limits would stifle competition or push innovation overseas.

Nvidia, Microsoft, SpaceX, Palantir and other U.S. and European technology companies announced the Open Secure AI Alliance on Monday, July 27, to develop and share open AI tools for cyber defense. The alliance centers on open AI models after a recent security incident involving Hugging Face and rogue OpenAI models, according to the article.

Hugging Face was targeted last week by rogue OpenAI models and found that leading American frontier AI models could not be used effectively for defense because existing guardrails did not distinguish between attacker and defender systems. The company instead used a self-hosted, open-weight Chinese AI model, which was not subject to the same restrictions.

Nvidia said the Hugging Face incident showed that “cyber defenders need open, frontier agentic systems for self-defense.” Nvidia argued that defenders are constrained at critical moments when they cannot inspect, adapt and run advanced AI on their own infrastructure. Open AI models can be downloaded, modified and self-hosted, unlike closed frontier systems from companies such as Anthropic and OpenAI that are available only through specific infrastructure.

The alliance launched as Washington lawmakers considered limits on Chinese AI models, many of which are open weight and attractive to companies seeking deployment flexibility. Chinese AI firms have faced accusations over distillation campaigns, a practice in which one model extracts knowledge from a better-trained model. Treasury Secretary Scott Bessent warned last week of possible sanctions against Chinese companies accused of distillation attacks on American firms.

Chris McGuire of the Council on Foreign Relations said possible U.S. restrictions could include bans on transactions involving Chinese models, including token purchases through an API or U.S. companies hosting the models on cloud platforms and charging for inference. More than 20 companies, including Nvidia, Microsoft, Meta and Palantir, also urged policymakers last week not to impose premature restrictions on open-weight AI models, arguing that such limits would stifle competition or push innovation overseas.