← Back to the feed

Okta buys AI security startup Permiso; source says for about $200M

TechCrunch ·

Okta has agreed to acquire AI identity-security startup Permiso Security, aiming to strengthen its ability to protect AI agents and other machine identities in corporate systems. The deal reflects a broader shift in identity security from checking users at login to monitoring the activity of people, applications and autonomous software after access has been granted.

Okta did not disclose terms, but a source told TechCrunch the transaction is worth just under $200 million and is almost entirely cash; it is expected to close in Okta’s third fiscal quarter of 2027. Permiso, founded by former FireEye executives and launched publicly in 2022, has raised about $29 million, developed cloud identity-threat detection tools and recently introduced SandyClaw to test AI-agent skills for malicious behaviour before deployment.

  • Okta is buying Permiso for reportedly just under $200 million.
  • The acquisition expands Okta’s AI-agent and machine-identity security capabilities.
  • Permiso has raised about $29 million since emerging from stealth.

Both sides, in good faith

The strongest fair case each way — we don't pick a winner.

The case for

Supporters would argue that buying Permiso is a timely way for Okta to address a genuine security gap as companies deploy AI agents and other machine identities with broad access to sensitive systems. Permiso’s monitoring and pre-deployment testing could complement Okta’s traditional access controls, helping customers detect harmful behaviour after credentials have been granted. They may also see a cash acquisition as a relatively focused investment in expertise built by experienced security practitioners.

The case against

Sceptics would argue that the reported near-$200 million price is substantial for a young company with limited disclosed commercial scale, especially in a fast-moving AI-security market where products and standards may change quickly. Integrating a specialist threat-detection platform into a large identity provider can be technically and organisationally difficult, and customers may prefer tools that remain independent or multi-vendor. They may also question whether AI-agent security is mature enough to justify a major acquisition before clearer demand and best practices have emerged.

AI Art Business Celebrity Companies Culture Entertainment Environment Science Technology

Read the full article at the source →