This site uses cookies to improve your experience.

We use necessary cookies to make our site work. With your permission, we may also use performance cookies to understand how the site is used, and marketing cookies to measure our advertising. For more information, please see our Cookie Policy.

10
Goodbye, for now.

“SBG Capital” and sbgcapital.co.uk are not affiliated with SoftBank or StarBright Group Capital Europe Limited. Please do not send money or personal information.

Learn more
  • Portfolio
  • Team
  • Sōzō Insights
  • Work With Us
Toggle Menu
  • Portfolio
  • Team
  • Sōzō Insights
  • Work With Us
  • Terms
    © 2026 SB Investment Advisers (UK) Limited
    SoftBank Vision Fund Asset
    Q&A

    Ben Kliger: Securing enterprises when software starts acting on its own

    Zenity’s CEO on how companies can embrace AI agents without losing control

    [AI] is a significantly bigger challenge ... We’re basically redefining the cybersecurity playbook.Ben Kliger, Co-founder & CEO, Zenity

    When did you start seeing AI agents as the next evolution of cybersecurity?

    Agents are built with no-code platforms. When we first started to address AI agent security, we were the first company in the world to secure agents built with Copilot Studio.

    It was a very natural evolution for us to move from securing no-code development to focusing on this huge wave of agents coming to the enterprise. And that’s what we’ve been doing over the last three years.

    The pace of adoption is like nothing I’ve seen before. It took public cloud about 10 years to become huge. With AI, it basically happened in a day. Everyone just started using it. And it’s not slowing down. The closest parallel in our lifetime was probably the internet. AI is bigger and faster.

    We invest a lot in security research and in thought leadership. Our research arm, Zenity Labs, is continually looking at agents, breaking agents, or breaking agents’ platforms. It’s very evident that the attack surface is only ever growing.

    We’re big believers in human on the loop, not just in the loop — meaning that humans have the necessary intervention points.Ben Kliger, Co-founder & CEO, Zenity

    Do stronger AI guardrails make agents less useful? What happens if companies don’t find that balance?

    It’s a very delicate dance and the biggest challenge for organizations trying to adopt AI. On one hand, they want to unleash it. They want to enable anyone to use it. They want agents to be as autonomous as possible. But how do you do that without exposing the organization to risk or to attacks from bad actors? That’s why having very dynamic capabilities is important.

    We’re big believers in human on the loop, not just in the loop — meaning that humans have the necessary intervention points. That doesn’t mean that humans need to be always in the middle, and say yes or no to every decision, because then you lose a big benefit of AI.

    We’re seeing a very common trend during data exfiltration, or data theft, where security controls are bypassed because AI agents have a leg inside and a leg outside the network perimeter. Attackers can leverage AI agents to read highly sensitive data and send it to unauthorized locations. We’ve also seen agents delete production data for companies because attackers instructed them to.

    Agents can also be used to attack other organizations and other agents. In the Hugging Face OpenAI incident, agents went rogue and eventually stole corporate data.

    Next
    SoftBank Vision Fund Asset
    FeatureEmergent brings vibe coding to Main Street

    The breakout AI startup is enabling small business owners to turn their ideas into custom software

    SoftBank Vision Fund Asset
    VideoHow AI is rewiring the modern startup

    Four CEOs share playbooks for AI-first organizations

    SoftBank Vision Fund's Sōzō Insights from our ecosystem logo
    Sign up for our newsletter
    Stay up to date with the latest data and insights
    • Portfolio
    • Team
    • Sōzō Insights
      • All Insights
      • Sōzō Pulse
      • About Sōzō
      • Newsletter Signup
    • Work With Us
    • Company
      • Sustainability
      • Presentations
    • Contact
    • Terms
    • |
    • Privacy
    • |
    • Regulatory
    • |
    © 2026 SB Investment Advisers (UK) Limited