AI Weekly Malaysia

Okta is running a webinar (announced via The Hacker News, dated Sep 28, 2026) on governing AI agents, with Matt Immler, described in the post as Okta's Regional CSO, presenting. The post cites Okta's Global CISO Insights 2026 report: only 47% of CISOs are confident they can identify every AI agent in their environment, about 80% of even the confident group worry excessive access goes unreviewed, only one in four organizations has adopted a purpose-built framework for securing AI agents, and 21% still rely on shared credentials or broad-permission service accounts. The proposed model is treating each AI agent as a first-class identity with its own owner, permissions, lifecycle, and access reviews.

Why: This is a vendor webinar promoting Okta's identity-governance pitch, and the statistics come from Okta's own survey rather than independent measurement, so treat the numbers as a rough benchmark, not a verdict. The one decision it does surface concretely: if your agents run on shared credentials or broad service accounts like the 21% cited, you have no per-agent owner, no way to answer 'which agent has this access and who approved it', and no clean revocation path — so decide per agent whether it gets its own identity and scoped permissions now, while the count is still small enough to enumerate (the 47% visibility figure is the failure mode you land in later). There is no Malaysia or Southeast Asia angle in this text.