93% of businesses say they understand AI risks "quite well" or "very well."
North Korea officially reported zero COVID-19 cases.
Gallagher just released its third annual AI Adoption and Risk Survey of 1,200+ global businesses.
Highlights:
- 63% have "operationalized" AI — whatever that means.
- 57% say AI errors and hallucinations are their top threat.
- Over half report not having the talent to manage the AI risks.
My take:
- We're in cybersecurity 2012. Everyone "understands" the risk. Almost no one is staffed, structured, or insured for it.
- Remember when every company was "confident" that they manage security risks well? Right before hiring their first CISO or getting breached? We're in that exact moment with AI risks. The confidence is based on not knowing what they don't know.
- Just like cyber insurance forced companies to adopt regular patching, MFA, and EDR, losses from currently uninsured AI runtime failures, model performance issues, and data supply chain attacks will kick off the cycle of real understanding.
