OpenAI Delays GPT-6.1 Over Security Flaws

OpenAI has announced it is delaying the release of GPT-6.1: internal reviews found the model too vulnerable to release publicly. Its response quality matches earlier versions — the problem lies specifically in defending against attacks.

Experts note the same trade-off between capability and security in models already available to users and businesses today. That means the risk isn't unique to a future release — it also applies to systems companies are plugging into their workflows right now. For business, this means impressive model capabilities don't guarantee safe handling of company data.

Check which AI tools are already connected in your company and who has access to their settings and logs. Confirm that employee permissions are properly restricted by role when working with these systems. Role-based access and a protected data perimeter for every workspace in 1Cifer keep company data under control even as underlying models change.