Security protects systems, data, and operations from threats, misuse, disruption, and unauthorized access. For critical AI applications, the risk is two-sided: AI can help detect threats, but malicious actors can also use it to launch cyberattacks and create destructive malware. section 11.16 section 1.2
Ransomware and critical infrastructure
Ransomware can remotely encrypt a victim’s files and demand payment for the decryption key. In 2015, a cyberattack forced Colonial Pipeline to shut down for a week; its CEO said the company paid a $4.4 million ransom amid the stakes of the disruption. section 11.16
Security concerns span both defenses and the systems being protected:
- AI can support threat detection and prevention in cybersecurity and public safety, while also introducing risks of misuse. section 1.2
- Smart agents incorporate security measures to protect sensitive data. section 4.2
