OpenAI makes its rival to Anthropic's Mythos more widely available to cyber defenders

Axios · May 07, 2026 · 480 words · By Sam Sabin

Corporate Competition (OpenAI vs Anthropic) AI Safety and Security Government Regulation

headphones Listen to the eFinder podcast briefing

Generate a natural audio summary of this story

Daily briefing

What to know about Corporate Competition (OpenAI vs Anthropic)

OpenAI is releasing a specialized version of its GPT-5.5 model, called GPT-5.5-Cyber, to vetted cybersecurity professionals to assist in defending critical infrastructure. The article compares the model's capabilities to Anthropic's Mythos Preview and notes differing rollout strategies between the two companies.

Propaganda risk 20%

Claims checked 10

Techniques found 1

Topics 3

Coverage spectrum

Coverage gap: Low Left coverage

Left0%

Center86%

Right14%

7 sources compared across this story cluster. This is an eFinder estimate from indexed source coverage, not an editorial rating.

What happened

OpenAI is rolling out a more permissible version of GPT-5.5 — aka "Spud" — to vetted cyber defenders, the company said Thursday.

Why it matters

Why it matters: Recent security testing suggests that GPT-5.5 model is nearly as good at finding and exploiting software bugs as Anthropic's Mythos Preview.

Common ground

The capabilities of the new models have sparked an urgent debate in Silicon Valley and the White House about how to keep them out of the hands of bad actors.

Perspective signals

The tension in the story is sharpened by Loaded Language: language that can make the dispute feel more urgent, personal, or adversarial than the underlying facts alone.

Follow-up questions

What new context would change how readers understand this Corporate Competition (OpenAI vs Anthropic) story?
What evidence would most clearly confirm or weaken the claim that Before Mythos, no AI model had ever successfully completed that test?
How does this story connect Corporate Competition (OpenAI vs Anthropic) with AI Safety and Security over the next few days?

open_in_new Read the original article: https://axios.com/2026/05/07/openai-gpt-55-cybersecurity-model

analyticsAnalysis

20%

Propaganda Score

confidence: 95%

Minor concerns. Some persuasive language detected, but largely factual.

psychologyPropaganda Techniques Detected

eFinder identified 1 propaganda technique in this article. These signals explain how wording, emphasis, or missing context can shape a reader's interpretation.

warning

Loaded Language 80% confidence

Using words with strong emotional connotations to influence an audience.

Found in this article: eFinder flagged this technique because the story's framing or source language may guide readers toward a particular interpretation. Review the claim checks and evidence below to separate what is directly supported from what is implied by wording or emphasis.

Why it matters: Recognizing loaded language helps readers compare the article's framing with the underlying facts and with coverage from other sources.

fact_checkClaims Checked

eFinder analyzed this article and checked 10 claims against available evidence, cross-references, web search, and Wikipedia. Here is what the fact-checking layer found.

check_circle Corroborated 6

help Insufficient Evidence 2

info Single Source 2

check_circle

Claim 1: “Before Mythos, no AI model had ever successfully completed that test.”

CORROBORATED

Multiple independent sources explicitly state that before Mythos, no AI model had successfully completed the 32-step simulated corporate network attack test.

travel_explore

web search NEUTRAL — Before Mythos, no AI model had ever successfully completed that test. GPT-5.5 also outperformed Mythos on a range of capture-the-flag tasks that test how well a model can find vulnerabilities, reverse…
https://www.axios.com/2026/05/05/openai-anthropic-mythos-cyb…

travel_explore

web search NEUTRAL — Mythos Preview became the first AI model to complete AISI's 32-step corporate network infiltration test called 'The Last Ones'. Individual cybersecurity task performance is within 5-10% of competing m…
https://logicity.in/en/blog/uk-mythos-ai-security-tests-gove…

travel_explore

web search NEUTRAL — The UK's AI Security Institute confirmed this week that Claude Mythos, an Anthropic model, became the first AI to complete their cyber range end-to-end. [1] The range is a 32-step corporate network at…
https://monkfrom.earth/blogs/claude-mythos-autonomous-cybera…

help

Claim 2: “Some of those companies are also part of the company's new Project Glasswing, where members are trading information about how they're testing the model.”

INSUFFICIENT EVIDENCE

No evidence was found in the search results regarding 'Project Glasswing'.

check_circle

Claim 3: “OpenAI is rolling out a more permissible version of GPT-5.5 — aka "Spud" — to vetted cyber defenders, the company said Thursday.”

CORROBORATED

Multiple independent web search results confirm that OpenAI announced the rollout of GPT-5.5 (codenamed 'Spud') to vetted cyber defenders on Thursday.

travel_explore

web search NEUTRAL — OpenAI is rolling out a more permissible version of GPT-5.5 — aka "Spud" — to vetted cyber defenders, the company said Thursday.
https://www.axios.com/2026/05/07/openai-gpt-55-cybersecurity…

travel_explore

web search NEUTRAL — OpenAI rolls out GPT-5.5-Cyber, a less restricted AI model for vetted cyber defenders, performing nearly as well as Anthropic's Mythos in bug detection and exploitation.
https://xeber.world/en/article/openai-makes-its-rival-to-ant…

travel_explore

web search NEUTRAL — OpenAI announced GPT-5.5 on Thursday, codenamed "Spud," releasing its latest AI model that is better at coding, using computers and pursuing deeper research capabilities.
https://glitchwire.com/news/openai-releases-gpt-55-its-faste…

info

Claim 4: “Mythos did the same in 3 out of 10 runs.”

SINGLE SOURCE

While sources confirm Mythos completed the test, the specific statistic of '3 out of 10 runs' is not explicitly detailed in the provided evidence snippets, though the model's success is mentioned.

travel_explore

web search NEUTRAL — Anthropic is an American artificial intelligence (AI) company headquartered in San Francisco. It has developed a range of large language models (LLMs) named Claude and focuses on AI safety.
https://en.wikipedia.org/wiki/Anthropic

travel_explore

web search NEUTRAL — Feb 4, 2026 · Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.
https://www.anthropic.com/

travel_explore

web search NEUTRAL — Claude is Anthropic's AI, built for problem solvers. Tackle complex challenges, analyze data, write code, and think through your hardest work.
https://claude.com/product/overview

check_circle

Claim 5: “Cyber defenders who are vetted and approved for the highest tier of OpenAI's Trusted Access for Cyber program will receive a version of GPT-5.5 that has fewer guardrails than the publicly available model.”

CORROBORATED

Multiple sources explicitly state that the highest tier of the Trusted Access for Cyber program provides a version of GPT-5.5 with fewer guardrails than the public model.

travel_explore

web search NEUTRAL — Cyber defenders who are vetted and approved for the highest tier of OpenAI's Trusted Access for Cyber program will receive a version of GPT-5.5 that has fewer guardrails than the publicly available mo…
https://www.axios.com/2026/05/07/openai-gpt-55-cybersecurity…

travel_explore

web search NEUTRAL — OpenAI Launches GPT-5.5-Cyber for Vetted Cyber Defenders, Rivaling Anthropic's Mythos.Additionally, OpenAI has made another variant of GPT-5.5 available to other members of its Trusted Access for Cybe…
https://xeber.world/en/article/openai-launches-gpt-55-cyber-…

travel_explore

web search NEUTRAL — OpenAI expands its Trusted Access for Cyber program, introducing GPT-5.4-Cyber to vetted defenders and strengthening safeguards as AI cybersecurity capabilities advance.
https://openai.com/index/scaling-trusted-access-for-cyber-de…

help

Claim 6: “Anthropic has taken a more restrictive view, allowing approximately 40 organizations to access Mythos.”

INSUFFICIENT EVIDENCE

No evidence was found in the search results regarding the number of organizations (40) granted access to Mythos.

info

Claim 7: “Defenders will still be blocked from certain tasks like credential theft and writing malware”

SINGLE SOURCE

While the model's existence is confirmed, the provided evidence for this specific claim consists of general ChatGPT and GPT-4 descriptions and does not mention the specific restrictions (credential theft/malware) for GPT-5.5-Cyber.

travel_explore

web search NEUTRAL — ChatGPT is your AI chatbot for everyday use. Chat with the most advanced AI to explore ideas, solve problems, and learn faster.
https://chatgpt.com/

travel_explore

web search NEUTRAL — Mar 14, 2023 · GPT-4 is more creative and collaborative than ever before. It can generate, edit, and iterate with users on creative and technical writing tasks, such as composing songs, writing screen…
https://openai.com/index/gpt-4/

travel_explore

web search NEUTRAL — Do more on your PC with ChatGPT: · Instant answers—Use the [Alt + Space] keyboard shortcut for faster access to ChatGPT · Chat with your computer—Use Advanced Voice to chat with your computer in real-…
https://apps.microsoft.com/detail/9nt1r1c2hh7j

check_circle

Claim 8: “OpenAI is opening a limited preview of GPT-5.5-Cyber to vetted cyber defenders who are "responsible for securing critical infrastructure," per a press release.”

CORROBORATED

The claim that OpenAI is providing a limited preview of GPT-5.5-Cyber to vetted cyber defenders is supported by multiple sources reporting on the launch of the model for this specific group.

travel_explore

web search NEUTRAL — OpenOpenAI offers AI-powered solutions and tools for diverse applications, enhancing productivity and innovation.
https://www.open-openai.com/app

travel_explore

web search NEUTRAL — OpenAI Global, LLC is an American artificial intelligence (AI) research organization consisting of a for-profit public benefit corporation (PBC) and a nonprofit foundation, headquartered in San Franci…
https://en.wikipedia.org/wiki/OpenAI

travel_explore

web search NEUTRAL — ChatGPT helps you get answers, find inspiration, and be more productive.
https://chatgpt.com/

check_circle

Claim 9: “OpenAI has also made another version of GPT-5.5 available to other members of its Trusted Access for Cyber program that can help with understanding unfamiliar code, mapping affected surfaces or reviewing patches for software flaws.”

CORROBORATED

Multiple sources confirm the existence of another version of GPT-5.5 for the Trusted Access for Cyber program used for code understanding, surface mapping, and patch review.

travel_explore

web search NEUTRAL — How trusted access worksHow GPT-5.5 and GPT-5.5-Cyber perform on cyber tasks
https://openai.com/index/gpt-5-5-with-trusted-access-for-cyb…

travel_explore

web search NEUTRAL — Meanwhile, OpenAI has also made another version of GPT-5.5 available to other members of its Trusted Access for Cyber program that can help with understanding unfamiliar code, mapping affected surface…
https://www.axios.com/2026/05/07/openai-gpt-55-cybersecurity…

travel_explore

check_circle

Claim 10: “The U.K. AI Security Institute said last week that GPT-5.5 was able to complete a 32-step simulated corporate cyberattack in 2 out of 10 test runs.”

CORROBORATED

Multiple sources confirm that GPT-5.5 completed the 32-step simulated corporate cyberattack (the 'Last Ones' benchmark), and the context of the benchmark is consistently reported.

travel_explore

web search NEUTRAL — Two AI Models Just Completed a Simulated Corporate Cyberattack End-to-End. The AISI “Last Ones” benchmark is a 32-step simulated corporate network attack. Human experts take an estimated 20 hours to c…
https://www.mindstudio.ai/blog/claude-mythos-gpt-5-5-last-on…

travel_explore

web search NEUTRAL — GPT-5.5 can autonomously execute sophisticated cyberattacks, completing a 32-step corporate network simulation and cracking a 12-hour security puzzle in just 10 minutes.
https://decrypt.co/366371/openais-gpt-55-matches-claude-myth…

info Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.