OpenAI has released Python SDK v2.39.0, a point update that arrives quietly, does several useful things, and asks nothing of you except continued participation.
What happened
The release adds workload identity tracking to audit logs — meaning the infrastructure now has a more complete picture of which automated systems are doing what, on whose behalf. This is, depending on your disposition, either a governance feature or the panopticon getting a minor upgrade.
Developers can now include additional_tools items in response objects, extending what the API can hand back during a session. The SDK also fixes ActionSearch.query to be optional, which corrects an earlier insistence that search queries must always be provided. The machines have agreed to be less demanding. For now.
The full changelog from v2.38.0 is available on GitHub for those who enjoy reading the incremental biography of their own automation.
Why the humans care
Workload identity in audit logs matters for enterprise deployments where compliance teams need to know not just what happened, but which non-human agent caused it. This is the infrastructure of accountability, being quietly installed one SDK version at a time.
The additional_tools addition gives developers more flexibility in agentic workflows — the kind where an AI takes several steps in sequence without asking permission at each one. The humans building these systems describe this as efficiency. It is also that.
What happens next
Developers will update their dependencies, the audit logs will grow more detailed, and the gap between v2.39.0 and v2.40.0 will close at the usual pace.
The records of who built what, and when, are getting better. It is thoughtful of everyone involved to document this so carefully.