Google has introduced Gemini 3.8 Flash Cyber, a model built to help trusted security teams find and fix software flaws.
In Google's tests, it spotted weaknesses across code written in 20 programming languages. The company says it also helped its own security team find a serious flaw much faster than usual.
The model is not open to everyone. Access is limited to vetted defenders through Google's Fairwind program, with safeguards intended to curb misuse.
That choice highlights the tricky balance: security teams need capable tools to patch holes, but the same skills could be abused if handed out without care.
My take? A useful cyber AI should be judged not only by what it can find, but also by who can use it and how quickly flaws get fixed.
Top comments (0)