
I just sat down with Nina Bambysheva from @Forbes to discuss AI in cybersecurity. She said something that most CISOs aren't ready to hear... Model capabilities are advancing rapidly, and individual weaknesses in live applications are now being chained into serious exploits. I saw this firsthand at @octane_security. Our product identified a UAF in Telegram’s desktop app and I handed it to a cyber model. The model found a second vulnerability and chained the two together, turning a crash-inducing dangling pointer into controlled reuse of the freed object in a stock client. My involvement was largely provisioning account access, and the investigation ran mostly on its own with very little steering. That ability to move from finding a bug to developing an exploit is changing what security teams need to prepare for. Defenders need to move faster. I’m proud to have our team’s research featured in Forbes. Thanks to @ninabambysheva for the thoughtful conversation, and to the Octane team doing this work every day. Full article in the comments 👇



















