Security Leaders Discuss AI Agent Controls and Cybersecurity Hiring Trends

Summary

Security Weekly hosts discuss AI agent permissions, identity oversight and incident preparation, alongside Cyber SN's cybersecurity jobs report through September 2026.

Key points

• The hosts argue that AI governance needs shared executive responsibility, with CISOs contributing security and risk expertise rather than owning the entire initiative.
• They warn that overprivileged AI agents can create security and privacy risks when organizations cannot audit their actions.
• Suggested controls include linking non-human identities to accountable people, limiting permissions, monitoring behavior, process isolation and just-in-time access.
• The speakers recommend short tabletop exercises to prepare for scenarios such as an agent locking administrators out of accounts.
• Discussing Cyber SN's report, the hosts say overall cybersecurity job postings fell about 26% year over year, while cyber data scientist postings grew 203%.
• The hosts report declines in several defensive roles and caution that governance must involve operational oversight rather than static policies.