Source A
OpenAI Has a New GPT-5.4-Cyber Model. Here's Why You Can't Use It
cnet.com
https://www.cnet.com/tech/services-and-software/openai-gpt-5-4-cyber-model-limited-access-cybersecurity-news/
Comparison
Winner: Source B is less manipulative
Source B appears less manipulative than Source A for this narrative.
Source B
Реверс-инжиниринг, поиск уязвимостей и анализ угроз — OpenAI обучила отдельную версию GPT-5.4 специально для киберзащитников
securitylab.ru
https://www.securitylab.ru/news/571618.php
Topics
Instant verdict
Less biased source: Source B
More emotional framing: Source A
More one-sided framing: Source A
Weaker evidence quality: Source A
More manipulative overall: Source A
Narrative conflict
Source A main narrative
Unlike Claude Mythos Preview, which Anthropic said is an entirely new model, OpenAI's GPT-5.4-Cyber is a fine-tuned version of its existing GPT-5.4 large language model.
Source B main narrative
The source links developments to economic constraints and resource interests.
Conflict summary
Stance contrast: emphasis on military escalation versus emphasis on economic factors.
Source A stance
Unlike Claude Mythos Preview, which Anthropic said is an entirely new model, OpenAI's GPT-5.4-Cyber is a fine-tuned version of its existing GPT-5.4 large language model.
Stance confidence: 69%
Source B stance
The source links developments to economic constraints and resource interests.
Stance confidence: 95%
Central stance contrast
Stance contrast: emphasis on military escalation versus emphasis on economic factors.
Why this pair fits comparison
- Candidate type: Closest similar
- Comparison quality: 52%
- Event overlap score: 26%
- Contrast score: 74%
- Contrast strength: Strong comparison
- Stance contrast strength: High
- Event overlap: Topical overlap is moderate. Issue framing and action profile overlap.
- Contrast signal: Stance contrast: emphasis on military escalation versus emphasis on economic factors.
Key claims and evidence
Key claims in source A
- Unlike Claude Mythos Preview, which Anthropic said is an entirely new model, OpenAI's GPT-5.4-Cyber is a fine-tuned version of its existing GPT-5.4 large language model.
- That was the logic behind Anthropic's Project Glasswing, announced last week.
- Instead, the company is doing a limited release to verified cybersecurity testers, according to a blog post shared on Tuesday.
- OpenAI uses the feedback from these testers for "understanding the differentiated benefits and risks of specific models, improving resilience to jailbreaks and other adversarial attacks, and improving defensive capabili…
Key claims in source B
- GPT-5.4-Cyber построили на базе GPT-5.4, но дополнительно дообучили для более свободной работы в легитимных сценариях кибербезопасности.
- Одобренные участники получат доступ к версиям существующих моделей, где будет меньше ограничений для учебных задач, защитного программирования и ответственных исследований уязвимостей.
- Одновременно злоумышленники тоже экспериментируют с новыми подходами, поэтому меры защиты, как считают в компании, нужно развивать вместе с ростом возможностей самих моделей.
- OpenAI объявила о расширении программы Trusted Access for Cyber и представила GPT-5.4-Cyber, новую версию модели для задач киберзащиты.
Text evidence
Evidence from source A
-
key claim
Unlike Claude Mythos Preview, which Anthropic said is an entirely new model, OpenAI's GPT-5.4-Cyber is a fine-tuned version of its existing GPT-5.4 large language model.
A key claim that anchors the narrative framing.
-
key claim
That was the logic behind Anthropic's Project Glasswing, announced last week.
A key claim that anchors the narrative framing.
-
causal claim
This is a common cybersecurity practice, one made all the more valuable and necessary because of AI.
Cause-effect claim shaping how events are explained.
-
omission candidate
GPT-5.4-Cyber построили на базе GPT-5.4, но дополнительно дообучили для более свободной работы в легитимных сценариях кибербезопасности.
Possible context gap: Source A gives less coverage to economic and resource context than Source B.
Evidence from source B
-
key claim
Одобренные участники получат доступ к версиям существующих моделей, где будет меньше ограничений для учебных задач, защитного программирования и ответственных исследований уязвимостей.
A key claim that anchors the narrative framing.
-
key claim
Одновременно злоумышленники тоже экспериментируют с новыми подходами, поэтому меры защиты, как считают в компании, нужно развивать вместе с ростом возможностей самих моделей.
A key claim that anchors the narrative framing.
-
emotional language
Реверс-инжиниринг, поиск уязвимостей и анализ угроз — OpenAI обучила отдельную версию GPT-5.4 специально для киберзащитников 18:04 / 15 апреля, 2026 2026-04-15T18:04:34+03:00 Alexander Anti…
Emotionally loaded wording that may amplify audience reaction.
-
evaluative label
GPT-5.4-Cyber построили на базе GPT-5.4, но дополнительно дообучили для более свободной работы в легитимных сценариях кибербезопасности.
Evaluative labeling that nudges a normative interpretation.
-
selective emphasis
Решение в OpenAI объясняют тем, что ИИ все активнее используют и защитники, и атакующие.
Possible selective emphasis on specific aspects of the story.
Bias/manipulation evidence
-
Source B · Framing effect
Решение в OpenAI объясняют тем, что ИИ все активнее используют и защитники, и атакующие.
Possible framing pattern: wording sets a specific interpretation frame rather than neutral description.
How score signals are formed
Bias score signal
Bias signal combines framing pressure, emotional wording, selective emphasis, and one-sided narrative markers.
Emotionality signal
Emotionality rises when evidence contains emotionally loaded wording and evaluative labels.
One-sidedness signal
One-sidedness rises when one frame dominates and alternative interpretations are weakly represented.
Evidence strength signal
Evidence strength rises with concrete claims, attributed statements, and verifiable contextual support.
Source A
35%
emotionality: 29 · one-sidedness: 35
Detected in Source A
appeal to fear
Source B
26%
emotionality: 25 · one-sidedness: 30
Detected in Source B
framing effect
Metrics
Framing differences
- Source A emotionality: 29/100 vs Source B: 25/100
- Source A one-sidedness: 35/100 vs Source B: 30/100
- Stance contrast: emphasis on military escalation versus emphasis on economic factors.
Possible omitted/downplayed context
- Source A pays less attention to economic and resource context than Source B.