OpenAI has announced the release of its latest cybersecurity model, GPT-5.4-Cyber, in response to concerns raised by competitors like Anthropic about the potential misuse of powerful AI models.

The move comes as Anthropic unveiled its Claude Mythos Preview with limited access due to fears over exploitation, prompting a wider industry discussion on AI safety and cybersecurity strategies.

OpenAI’s strategy revolves around three key pillars: controlled access through validation systems, iterative deployment processes, and investment in software security. The company is also expanding its existing guardrails while acknowledging the need for more advanced protections as models evolve.

The announcement reflects a nuanced approach, seeking to reassure users that current safeguards are sufficient for broad model deployment but hinting at future needs for more robust measures. This strategy is part of OpenAI’s broader security efforts, including initiatives like Codex Security and cybersecurity grants programs.