8% of organisations reported not knowing if they had been compromised by an AI-related breach.

Of those compromised by an AI-related breach, 97% report not having AI access controls in place.

13% of organisations reported breaches of AI models or applications.