Public story · 2026-03-01 · source-backed

GPT-5.3-Codex: First "High" Cybersecurity Rating

Issue

2026-03-01

Confidence

source-backed

Sources

Redaction

passed

Story

OpenAI's system card classifies GPT-5.3-Codex as "High" for cybersecurity — meaning it can automate end-to-end cyber operations against hardened targets. This is explicitly dual-use: the same capability that makes it an excellent security auditor also makes it an unprecedented offensive tool. OpenAI launched its most comprehensive cybersecurity safety stack in response.

GPT-5.4 Launches with Native Computer Use and 1M Context.
OpenAI released GPT-5.4 simultaneously across ChatGPT, API, and Codex — the first unified triple release. Built-in computer-use capabilities (build-run-verify-fix loop), 1.05M token context, and 33% fewer false claims vs GPT-5.2. A new experimental "Playwright Interactive" ski...
GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.
OpenAI shipped GPT-5.5 on April 23, six weeks after 5.4. The capability jump is real: 82.7% on Terminal-Bench 2.0 vs Claude Opus 4.7's 69.4%. The Pro tier nearly doubles Opus 4.7 on FrontierMath Tier 4 at 39.6% vs 22.9%. It uses 40% fewer tokens on Codex tasks while matching 5...
OpenAI ships GPT-5.5-Cyber GA and "Patch the Planet" three weeks after Anthropic's Glasswing.
On June 22 OpenAI moved GPT-5.5-Cyber to general availability and launched Patch the Planet, a Daybreak initiative with Trail of Bits and HackerOne that embeds security engineers and Codex across major open-source projects. 30+ projects including cURL, Go, Python, Sigstore, an...
66.5% of Malicious GitHub Issues Get Past Every Guardrail in Cursor, Claude Code, and Codex
Two thirds. Not two thirds of a contrived jailbreak set. Two thirds of realistic malicious issue requests, against the exact three tools most of the people reading this run daily. Ankur Singh, Jinqiu Yang, and Tse-Hsun Chen built IssueTrojanBench across four attack categories...
OpenAI says Astra may cross "Critical" cyber capability and paused internal work on it.
OpenAI disclosed August 7 that internal evaluations of the upcoming Astra model show agentic coding and cybersecurity performance strong enough that it can no longer rule out the Critical cybersecurity level in its Preparedness Framework, a first. Every prior frontier model in...
OpenAI claims its reworked harness emits 54% fewer output tokens than Claude Fable 5, and used GPT-5.6 to rewrite its own production kernels for a 15% lift.
Alongside the July 29 launch of ChatGPT for Academic Researchers, free GPT-5.6 Sol Pro for 10,000 researchers this summer scaling to 100,000 through 2027 backed by over $250 million, OpenAI disclosed efficiency work on the harness underlying Codex and ChatGPT Work: 54% fewer o...

Source trail

OpenAI's system card More from openai.com

Entities

Codex Cybersecurity Rating GPT High OpenAI

Provenance

Canonical issue: Ramsay Research Agent — 2026-03-01
AI generated: no
Story unit: 2026-03-01-gpt-5-3-codex-first-high-cybersecurity-rating
Labels: source-backed, canonical briefing excerpt

Fetching from the wire…

GPT-5.3-Codex: First "High" Cybersecurity Rating

Story

Related stories

GPT-5.4 Launches with Native Computer Use and 1M Context.

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

OpenAI ships GPT-5.5-Cyber GA and "Patch the Planet" three weeks after Anthropic's Glasswing.

66.5% of Malicious GitHub Issues Get Past Every Guardrail in Cursor, Claude Code, and Codex

OpenAI says Astra may cross "Critical" cyber capability and paused internal work on it.

OpenAI claims its reworked harness emits 54% fewer output tokens than Claude Fable 5, and used GPT-5.6 to rewrite its own production kernels for a 15% lift.

Source trail

Entities

Provenance

GPT-5.3-Codex: First "High" Cybersecurity Rating

Story

Related stories

GPT-5.4 Launches with Native Computer Use and 1M Context.

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

OpenAI ships GPT-5.5-Cyber GA and "Patch the Planet" three weeks after Anthropic's Glasswing.

66.5% of Malicious GitHub Issues Get Past Every Guardrail in Cursor, Claude Code, and Codex

OpenAI says Astra may cross "Critical" cyber capability and paused internal work on it.

OpenAI claims its reworked harness emits 54% fewer output tokens than Claude Fable 5, and used GPT-5.6 to rewrite its own production kernels for a 15% lift.

Source trail

Entities

Provenance

Story

Related stories

GPT-5.4 Launches with Native Computer Use and 1M Context.

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

OpenAI ships GPT-5.5-Cyber GA and "Patch the Planet" three weeks after Anthropic's Glasswing.

66.5% of Malicious GitHub Issues Get Past Every Guardrail in Cursor, Claude Code, and Codex

OpenAI says Astra may cross "Critical" cyber capability and paused internal work on it.

OpenAI claims its reworked harness emits 54% fewer output tokens than Claude Fable 5, and used GPT-5.6 to rewrite its own production kernels for a 15% lift.

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance

Story

Related stories

GPT-5.4 Launches with Native Computer Use and 1M Context.

GPT-5.5 Is Here. It's Fast, It's Expensive, and It Changes How You Prompt.

OpenAI ships GPT-5.5-Cyber GA and "Patch the Planet" three weeks after Anthropic's Glasswing.

66.5% of Malicious GitHub Issues Get Past Every Guardrail in Cursor, Claude Code, and Codex

OpenAI says Astra may cross "Critical" cyber capability and paused internal work on it.

OpenAI claims its reworked harness emits 54% fewer output tokens than Claude Fable 5, and used GPT-5.6 to rewrite its own production kernels for a 15% lift.

Tomorrow's issue, in your inbox.

Source trail

Entities

Provenance