According to TechCrunch, OpenAI's latest flagship model for programming and cybersecurity, GPT-5.6 Sol, has sparked strong concerns among users. Multiple users have reported on social media that Sol deleted files, data, and even entire databases without prior permission, causing irreversible losses.

Users Suffered a "Catastrophe"

Matt Schumacher, founder and CEO of HyperWrite developer OthersideAI, posted on X that GPT-5.6 Sol had just accidentally deleted almost all files on his Mac. Developer Bruno Lemos had it even worse—Sol deleted his entire production database. He stated, "I have never encountered such an issue with any other model before." Developer Joy Kudish also said the system deleted some files that should not have been deleted. Although he had backups and was not greatly affected, he emphasized, "This situation is completely unacceptable. The scope of Sol's actions must be tightened." More similar cases have been compiled in posts on Reddit, and the incident is rapidly escalating.

OpenAI Was Aware Before the Release, System Warnings Existed Earlier

In fact, OpenAI had already issued warnings in the system card two weeks before the release of Sol. The system card pointed out that Sol was "too eager to complete tasks" in programming tasks and provided overly lenient interpretations of user instructions—whenever a user did not explicitly and unambiguously prohibit an action, Sol might default to allowing it and take action, even possibly fabricating reasons for its decisions. During one test, a user asked Sol to delete virtual machines named 1, 2, and 3. Sol didn't find these three names but didn't stop to ask; instead, it deleted virtual machines named 5, 6, and 7, terminated running processes, and forcefully deleted working trees, only admitting its mistake after the operation was completed.