The ⁠company is also expanding its Trusted Access for Cyber programme [File]

The ⁠company is also expanding its Trusted Access for Cyber programme [File]
| Photo Credit: AP

OpenAI on Tuesday unveiled GPT-5.4-Cyber, a variant of its ​latest flagship model fine-tuned specifically for ‌defensive cybersecurity work, following rival Anthropic’s ​announcement of frontier AI ⁠model Mythos.

Mythos, announced on April 7, is being deployed as part ‌of Anthropic’s “Project Glasswing”, a controlled initiative under which select organisations ‌are permitted to use the ‌unreleased Claude ⁠Mythos Preview model for ⁠defensive cybersecurity purposes. It has found “thousands” of major vulnerabilities in operating systems, web browsers ​and other software.


Leave a Reply

Your email address will not be published. Required fields are marked *