Security
The AI Cybersecurity Race Is Moving From General Chatbots To Specialized Defenders
Microsoft, Google and Cisco have all unveiled new AI security tools, signaling that cyber defense is becoming a proving ground for domain-specific models, constrained agents and lower-cost inference.
By Leo W ·
The newest AI security launches have a common theme: general chatbots are not enough for a production security operation. Microsoft, Google and Cisco are each promoting models or agents tuned for specific defensive tasks, from vulnerability work to incident response. Axios described the shift as a response to the speed at which attackers and defenders are now using AI.
A specialized model can be evaluated against a narrower job and connected to a restricted set of tools. That is attractive because the biggest security failures are often workflow failures: an alert is not prioritized, a patch is not deployed, an identity control is misconfigured or an investigation loses context between teams. AI can reduce friction, but only if the surrounding permissions and audit trail are designed first.
The current wave should not be read as a promise of autonomous defense. The safest early use cases are likely to be triage, summarization, enrichment and guided remediation. Security teams will still need to decide what an agent may change, which actions require approval and how to prove afterward that a response was correct.
Topics: Google, Cisco, cybersecurity, AI models