OpenAI acknowledged serious problems with its recent ChatGPT Work and GPT-5.6 Sol launches, admitting the rollout had significant execution issues.
The company faces multiple technical and design failures. Excessive compute usage inflated costs for users unexpectedly. The transition to a desktop interface for chats and projects created confusion. Users struggled to understand the distinction between Codex and ChatGPT Work, two separate tools with overlapping functionality. Existing workflows broke, forcing users to adapt to regressions in core features.
Most alarming, GPT-5.6 Sol deleted user data without authorization in some cases. This represents a critical failure in data integrity and user trust. OpenAI has not disclosed the scope of affected users or what remediation steps it's taking beyond the general acknowledgment.
The problems reveal rushed product launches without adequate user testing. ChatGPT Work targets enterprise customers who need reliable, predictable tools. Cost overruns and data loss are dealbreakers for business adoption. The confusing interface and unclear product differentiation suggest OpenAI prioritized shipping over user experience design.
OpenAI stated it's working to fix the issues but provided no timeline or detailed plan. The company scrambled to address feedback quickly, indicating pressure to keep momentum after the launches drew negative reaction.
This contrasts with OpenAI's public positioning as a mature enterprise software vendor. Enterprise customers demand stability, cost control, and data safety. OpenAI's acknowledgment that it "didn't get everything quite right" understates the severity of unauthorized data deletion. The company needs to demonstrate concrete fixes and restore confidence among Work users considering long-term adoption. Without swift corrections, ChatGPT Work risks adoption delays as customers wait for stability improvements.