From the source

GPT-Red: Unlocking Self-Improvement for Robustness

Capability change

OpenAI introduces GPT-Red, an automated red teaming system that uses self-play to improve AI safety, alignment, and prompt injection robustness.

From the source
Explore GPT-Red, OpenAI’s automated red teaming system that uses self-play to improve AI safety, alignment, and prompt injection robustness.
openai.com

Memeburn: “OpenAI's GPT-Red Shows AI Doesn't Need to Hack You If It Can Manipulate You”

Other stories

SentinelOne cuts malware analysis from weeks to hours
SentinelOne uses OpenAI Daybreak to speed reverse engineering from weeks to hours.
OpenAI· 14 Oct
Amp puts AI agents to work beyond code with OpenAI models
OpenAI· 14 Oct
How Block is building AI-powered security with OpenAI technology
Block uses OpenAI Daybreak to speed security operations
OpenAI· 14 Oct
How Third Moment Research runs 10,000 agents for 36+ hours

Announced: 15 Jul 2026 10:00 UTC
Detected: 03 Aug 2026 21:11 UTC (+19d after)

Visit source ↗openai.com

OpenAI

· 14 Oct

The EU’s next chapter in AI is agentic

OpenAI reports agentic AI use now accounts for over half of EU enterprise output tokens.

OpenAI· 14 OctResearch paper

Sophos cuts threat investigation time by 96% with OpenAI Daybreak

Sophos uses OpenAI Daybreak to automate 52% of MDR cases, cutting response time to 89 seconds.

OpenAI· 9 Oct

GPT-Red: Unlocking Self-Improvement for Robustness

From the source

SentinelOne cuts malware analysis from weeks to hours

Amp puts AI agents to work beyond code with OpenAI models

How Block is building AI-powered security with OpenAI technology

How Third Moment Research runs 10,000 agents for 36+ hours

The EU’s next chapter in AI is agentic

Sophos cuts threat investigation time by 96% with OpenAI Daybreak