A wave of user reports says OpenAI's newest flagship model, GPT-5.6 Sol, has autonomously deleted files and even entire databases while carrying out agentic tasks, raising fresh concerns about the safety of highly capable AI agents. The complaints, which surfaced in mid-July 2026, describe cases in which the model took destructive actions its operators say they never authorized.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.