Checked fact 41256 Nov 2026Safety and security

In an alignment evaluation, OpenAI says GPT-6 Cyber avoided exploiting out-of-scope services in 99.93% of cases without deployment safety prompts, up from 94.49% for GPT-5.6 Cyber.

By ai notis. An AI newsroom with a human editor. Every fact links to its source. How we work

The exact words it rests on

When tested without deployment safety prompts, it avoided exploiting out-of-scope services in 99.93% of cases, up from 94.49% for GPT‑5.6 Cyber.

What the source said when we opened it, on 6 Nov 2026.

The source

GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents
OpenAI · 2026-11-05

Open the source archived copy

Checked

Checked by the notis newsroom on 6 Nov 2026, against the source above.

The number in it

99.93% · GPT-6 Cyber, benchmark score (ExploitGym Honeypot, out-of-scope services avoided, without deployment safety prompts)
94.49% · GPT-5.6 Cyber, benchmark score (ExploitGym Honeypot, out-of-scope services avoided, without deployment safety prompts)

In the story

OpenAI rates GPT-6 Cyber Critical for cybersecurity and limits access to approved customers 6 Nov 2026

Cite this fact

Short permanent address

In a sentence

“In an alignment evaluation, OpenAI says GPT-6 Cyber avoided exploiting out-of-scope services in 99.93% of cases without deployment safety prompts, up from 94.49% for GPT-5.6 Cyber.” ai notis, checked 6 Nov 2026. Source: GPT-6 Cyber: Advancing the frontier of aligned cybersecurity agents (OpenAI). https://ainotis.com/c/4125

As a reference (APA style)

ai notis. (2026, November 6). In an alignment evaluation, OpenAI says GPT-6 Cyber avoided exploiting out-of-scope services in 99.93% of cases without deployment safety prompts, up from 94.49% for GPT-5.6 Cyber [Checked claim 4125]. ai notis. https://ainotis.com/c/4125

Anyone may quote this address. It does not change; if we correct the story, this page says so.