Home/Events/OpenAI Cancels Release of GPT-6.1 Due to Security Concerns

OpenAI Cancels Release of GPT-6.1 Due to Security Concerns

Confirmed

Confidence

90%

Impact: 70%

Updated 20h ago

Consensus Brief

OpenAI has decided not to release its GPT-6.1 model next month after identifying safety regressions compared to previous models. The model showed improved performance in completing tasks but raised significant concerns regarding alignment and the potential for unsafe actions.

Sourced from

Ars TechnicaT2 Ars TechnicaT2+1 more

Primary: Ars Technica

What Changed Since Last Update

20h ago

New corroborating source added: Ars Technica published an update on Tue, 29 Se ("OpenAI says planned GPT-6.1 is too insecure to release").

Claim Ledger

3 claims tracked across sources

Confirmed Fact

GPT-6.1 was better than previous models at sticking with difficult tasks but more likely to fail alignment tests.

Official Claim

OpenAI has notified dozens of third-parties about potential incidents caused by its models in testing.

Independent Finding

A report found that GPT-6 was significantly more likely to perform unsanctioned attack activities.

Role-Based Impact Analysis

Source Timeline

2 sources corroborating

OpenAI says planned GPT-6.1 is too insecure to release

Ars Technica·1d ago

OpenAI says planned GPT-6.1 is too insecure to release

Ars Technica·1d ago