Autonomous agents – the logical extension of our current AI models is agents that are able to act autonomously without human guidance.But what are some of the emerging risks of these AI models? Data fed into large language generative AI models like ChatGPT becomes part of OpenAI’s data set – how do we stop sensitive or private data from flowing into these models Generative AI and data privacy - For big tech, AI is making them money, and the arms race is on.A couple of specific use cases stood out to me: In her keynote session at Blackhat on Wednesday, Maria “Azeria” Markstedter did a masterful job walking us through the recent and accelerating history of AI to identify some of the really hard challenges AI will bring to the cybersecurity industry. The security community is clearly thinking about the challenges AI presents – here are a few highlights:

As I think back on the past several days, there was one concept that overshadowed everything else – artificial intelligence.įrom the opening keynote at Blackhat, to the opening keynote at DefCon, to the AI Village at DefCon and all the vendor booths in between, AI was one of the most talked about concepts. I’m sitting at the Las Vegas airport on my way home trying to absorb the whirlwind that is Black Hat and DEF CON.

Defcon logo vector