Brief IA

OpenAI Restricts Access to Its Cybersecurity AI Following Anthropic

🤖 Models & LLM·Tom Levy·

OpenAI Restricts Access to Its Cybersecurity AI Following Anthropic

OpenAI Restricts Access to Its Cybersecurity AI Following Anthropic
Key Takeaways
1OpenAI is developing an advanced cybersecurity AI, accessible only to a select group of companies.
2Anthropic has also restricted its Mythos Preview model to selected tech and security companies.
3OpenAI's pilot program, launched in February, offers $10 million in API credits for defensive security work.
💡Why it mattersThese restrictions highlight the growing concerns regarding the security and ethics of powerful AIs in the field of cybersecurity.
Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisis ton rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

📄
Full Analysis

OpenAI Restricts Access to Its Cybersecurity AI

OpenAI, one of the leaders in artificial intelligence, is currently developing an advanced AI model in cybersecurity. However, unlike a broad rollout, this model will only be accessible to a limited number of companies. This strategy is reminiscent of that adopted by Anthropic, another company in the sector, which has also restricted access to its Mythos Preview model. The latter is reserved for certain technology and security companies due to its highly powerful hacking capabilities.

An Exclusive Pilot Program

In February, OpenAI had already initiated a pilot program titled "Trusted Access for Cyber." This program was launched following the release of GPT-5.3-Codex, considered the company's most advanced cybersecurity model to date. Participants in this program benefit from privileged access to powerful models for defensive security missions. This program is supported by API credits worth $10 million.

Uncertainties About Broader Availability

It remains to be seen whether OpenAI will decide to make this new cybersecurity model more widely accessible in the future. For its part, Anthropic has clearly stated that its Mythos Preview model will not be available to the general public until adequate security measures are in place. This cautious approach underscores the importance placed on security and ethics in the development and deployment of powerful AI technologies.

Brief IA — L'actualité IA en français

L'essentiel de l'actualité de l'intelligence artificielle, décrypté et expliqué chaque jour.