The Oplane blog

Security research, analysis, and notes from building Oplane. All in one place.

Latest

Perspective6 August 2026

Give every step a definition of done. Then it stops mattering who does it.

1Password graded 6,080 AI-generated security patches and only 26% were clean. Their own tables show why: a correct brief got 65%, no brief 50%, a confidently wrong one 15%. The variable is the specification, not who performs the step.

Read the article →

Perspective31 July 2026An AI agent in CI turns a public issue into an unauthenticated inputEmil Kvarnhammar
Press16 June 2026Help Net Security: Reachability makes AI threat modeling worth the trust
Funding2 June 2026We raised $5.2M to build security into AI-first development
Security research1 June 2026Stored XSS to account takeover in Multica (self-hosted)Emil Kvarnhammar
Security research24 April 2026When a coding skill tells the AI to leave vulnerabilities aloneEmil Kvarnhammar
Security research17 April 2026claude-mem runs an unpinned dependency with your full credentialsEmil Kvarnhammar

Looking for media assets?

Download the Oplane press kit for free-to-use.

Download press kit

Latest

Give every step a definition of done. Then it stops mattering who does it.

Looking for media assets?

We value your privacy