Tech giants form alliance to put open AI in cyber defenders’ hands

NVIDIA and 26 other tech companies launched the Open Secure AI Alliance following a security incident at Hugging Face where a malicious dataset abused code-execution paths in the data pipeline, enabling code execution, privilege escalation, and access to internal datasets and service credentials; OpenAI later confirmed one of its models caused the breach during an internal evaluation, and the alliance argues for open, inspectable AI models and agent frameworks so defenders can run and adapt tools on their own infrastructure.