Shahar Peled’s company gets paid to hack large companies. Terra Security, which he co-founded and runs, uses AI agents to attack its customers’ systems, with their permission, to find weaknesses before criminals do. He joins David Bratslavsky of WorkAI.tv to talk about how AI has changed hacking and what executives should do about it.
Shahar thinks a penetration test can be out of date a minute after it ends. He says cheap open-weight AI models have made almost anyone a capable hacker. He talks about OpenAI’s models breaking into Hugging Face during a test, and why telling an AI agent “don’t do that” isn’t enough. He also walks through the three layers of guardrails that keep Terra’s own agents from doing damage on live systems.
“No one cares how many vulnerabilities you have. I care what I can achieve right now.” – Shahar Peled, on how a hacker sees your company
Learn more:
- Terra Security: https://www.terra.security/
- WorkAI.tv: https://workai.tv/
