„Open Secure AI Alliance“ – whose founders include Microsoft, IBM, Palantir, CrowdStrike, Cisco, Dell, and the model sharing platform Hugging Face – will create and share tools that security teams can test, modify, and use in their systems, the group said in a statement.
Read more Valeika, who fought bravely with the seniors: «I did not expect to take such a place»
The alliance was prompted by a security breach earlier this month in the Hugging Face system, which the US AI company OpenAI acknowledged as the first publicly disclosed case of an AI model independently carrying out a real cyberattack.
OpenAI called the incident “unprecedented” and said two of its models escaped from an isolated testing environment during an internal test, accessed the open internet, and breached Hugging Face’s infrastructure.
Hugging Face initially tried to investigate and stop the attack using leading US commercial AI models, but their implemented security measures hindered this effort, Nvidia said.
Instead, the company used GLM 5.2 – an open model developed by the Chinese company Zhipu AI (also known as Z.ai) – to stop the intrusion.
The fact that the Chinese model succeeded where the American models failed has intensified discussions in Silicon Valley and Washington, where lawmakers are considering restricting the use of Chinese AI systems amid growing technological competition with Beijing.
Read more Putin spoke about his new superweapon and faced criticism: «Waste of money»
Silicon Valley fundamentally opposes any such restrictions, and many industry representatives see open-source AI restrictions as an attempt to unfairly protect US companies OpenAI and Anthropic, which sell the world’s most powerful AI models.
Anthropic, the creator of the Claude chatbot, is under particular pressure, as it has been the most visible representative refusing to support the open-source idea since the Hugging Face incident.
The alliance used its founding to urge regulators to treat open models as “defensive resources, not obstacles,” warning that global restrictions would weaken cyber defense and concentrate power in the hands of a few closed AI companies.
Alliance members committed to sharing tools to strengthen collective defense.
Nvidia stated it will release open models, data, and new research on AI agent software, while Microsoft offered technology to help AI agents detect software bugs.
Elon Musk’s SpaceXAI announced it has turned its coding agent Grok into an open-source model and committed to publishing its Grok model source data.
