Mistral open-sources AI model despite warning it could aid cyberattacks

Mistral AI has released the open weights of its trillion-parameter Mistral Large 4 model, despite warning the model could help attackers carry out real cyberattacks, CSO Online reports. The company said it monitored cybersecurity-related use during the model’s public preview and observed it refusing black-box attacks against live targets, but still saw potential for uplift to conduct real attacks.