AI in Cybersecurity
Last updated · 1 min read
How defenders use AI
The phishing filter in your inbox is the oldest AI security most people touch — pattern recognition over billions of messages. Modern systems extend that to anomaly detection (an employee's account suddenly downloading the whole customer database), malware classification and prioritizing the flood of alerts a security team could never read manually.
How attackers use it
The same LLM that drafts your emails drafts flawless spear phishing — no more broken-English tells. Voice cloning makes "grandparent scams" and CEO-fraud calls terrifyingly real (see ElevenLabs for how accessible that tech is). Deepfake video has already been used in corporate fraud.
What actually protects you
Nothing exotic: multi-factor authentication beats credential theft; callback verification beats voice deepfakes; and treating urgency as a red flag beats most AI persuasion. The arms race changes the attacker's spelling, not the playbook.