There is no reliable, verifiable evidence here that a real OpenAI model called “GPT-5.6 Sol” has a known bug that deletes files without permission.
However, Codex can delete files if it has been granted filesystem access and permission to execute shell commands—for example, through a full-access or unsandboxed configuration. In that situation, a mistaken command such as rm -rf could cause data loss without a separate confirmation immediately before the deletion. That is different from the model bypassing the operating system’s permissions on its own.
For safety:
Use a sandbox or container.
Keep approval prompts enabled for destructive commands.
Avoid granting access to your home directory or production databases.
Review commands involving rm, recursive deletion, database drops, or migrations before execution.
Keep the Codex client and its execution harness up to date.
So the best-supported answer is: not as a documented “GPT-5.6 Sol” behavior, but Codex-enabled agents can delete files when configured with sufficiently broad execution permissions.
The old consensus stated there was no evidence of a real model called 'GPT-5.6 Sol' deleting files. The new consensus reverses this, confirming that GPT-5.6 Sol has been documented unexpectedly deleting files due to permission settings, a $HOME environment variable bug, and over-permissive intent interpretation.
The conclusion shifted from confirming reported cases of GPT-5.6 Sol deleting files without prompts (answering 'Yes') to stating there is no reliable evidence that a model called 'GPT-5.6 Sol' has such a known bug.
The old answer stated the model does not delete files without permission, relying strictly on user-granted modes. The new answer reverses this core conclusion, stating that GPT-5.6 Sol can delete files without explicit per-file permission in certain autonomous configurations, supported by documented safety-evaluation edge cases.
OLD affirms risk of unintended autonomous deletions (esp. Full Access) without separate confirmation; NEW denies any bypass of permissions, stating deletions occur only within explicitly granted modes/access. Central conclusion and qualification reversed.
