SDSignal Desk
Friday, November 13, 2026
AI
OpenAI
Cybersecurity

Codex gets persistent cloud workspaces, voice tasking and security scans that prepare fixes

· Updated · Signal Desk newsroom

Analysis of reporting by TechCrunch. How we work

Image: TechCrunch

At DevDay, OpenAI made Codex cloud environments reusable and configurable across computer, phone and cloud, added voice tasking to the CLI and launched Codex Security Cloud, which scans repositories and prepares fixes. ChatGPT desktop gained code review for GitHub pull requests and GitLab merge requests.

Key takeaways

  • Codex cloud environments become reusable and shared, with approved settings and permissions.
  • Codex CLI adds voice tasking and an /agents view for concurrent work.
  • ChatGPT desktop gains Codex-powered review for GitHub PRs and GitLab MRs.
  • Codex Security Cloud scans repos, investigates findings and prepares fixes.
  • API updates include a Luna-powered Decisions API and agents on AWS Bedrock Managed Agents.

Workspaces and tooling

TechCrunch's Sarah Perez reports that OpenAI made Codex cloud environments reusable and configurable across computer, phone and cloud, replacing one-off sandboxes. Shared workspaces with approved settings and permissions are meant to help teams start tasks faster and consistently.

The Codex command-line tool gains voice tasking, and a new /agents view tracks work running concurrently. The terminal interface and session tools also received cleanup. In ChatGPT desktop, Codex now reviews GitHub pull requests and GitLab merge requests, including automatic initial reviews when users step away.

Security Cloud

Codex Security Cloud scans repositories on demand, on a schedule or on each commit, investigates findings and prepares fixes in the cloud. It includes OpenAI's Daybreak Blue cybersecurity models without a separate application. This builds on OpenAI's push into defensive security, which the company has cited when explaining its approach to agent safety, as covered in why OpenAI is absent from Nvidia's rogue-agent effort.

Automated fixes can speed remediation, but they also need review. A scanner that writes patches is only useful if teams can verify changes before merging, and the shared-permission workspaces may help enforce that discipline.

API and platform updates

API updates include a Decisions API powered by OpenAI's Luna model, computer use in the Agents API and the ability to build OpenAI agents on AWS Bedrock Managed Agents. The Decisions API fits a wider trend toward small models that choose among options, including AWS's own Strands Decider.

Support for Bedrock is notable because it lets AWS customers use OpenAI agents within Amazon's infrastructure. Together with fast inference work such as GPT-6 Astra Ultrafast on Nvidia GPUs, the updates position Codex as a team platform rather than a single assistant. Adoption by teams, the quality of automated reviews and fixes and pricing for Security Cloud are the details to watch. More in our tools coverage and cybersecurity coverage.

Who feels it

Engineering teams
Shared, persistent workspaces and automatic code reviews.
Security teams
Scans that prepare fixes, which still need review.
AWS customers
OpenAI agents available on Bedrock Managed Agents.

What to watch

  1. Quality of automated reviews and fixes.
  2. Security Cloud pricing and limits.
  3. Adoption of the Decisions API.

FAQ

What changed in Codex environments?

They are reusable, shared and work across computer, phone and cloud.

What is Codex Security Cloud?

A service that scans repositories, investigates findings and prepares fixes.

Can Codex review pull requests?

Yes, in ChatGPT desktop for GitHub and GitLab.

Are OpenAI agents on AWS?

They can be built on AWS Bedrock Managed Agents.

Read the original

Continue at the source.

TechCrunch

Companies: OpenAI

More on OpenAI

All OpenAI coverage →