Open AI expands the Daybreak security initiative and launches GPT-5.6-Cyber with capabilities that exceed the standard model

Open AI has announced the expansion of its security initiative, Daybreak, with the introduction of a specialized model called GPT-5.6-Cyber, designed for vulnerability research and authorized security testing. This move comes at a time when concerns are growing about the use of artificial intelligence in large-scale cyber attacks, and puts the new model in the hands of trusted defenders before attackers can deploy equivalent offensive capabilities.

Model with different specifications

According to Open AI's internal assessment, GPT-5.6-Cyber completed 95 percent of advanced cyber security requests, compared to only 1.5 percent for the standard model GPT-5.6 Sol. The large gap reflects the design decision: the new model has relaxed the usual restrictions that prevent the generation of exploitable instructions, while maintaining barriers to prevent misuse by unauthorized parties.

Discovery of real vulnerabilities in V8

The model was used in a study of widely used software, and the research found two previously unknown vulnerabilities in the V8 engine that powers the Chrome browser. These vulnerabilities can be chained to corrupt memory and escape the browser's sandbox, a finding that is practical evidence of the model's ability to contribute to both offensive and defensive security research.

Readiness framework: high threshold without critical

Within the Preparedness Framework established by Open AI in 2023, both GPT-5.6 Sol and GPT-5.6-Cyber are classified at the high cyber capability threshold, without reaching the critical level. This classification means that the models show significant efficiency in complex cyber security tasks, but have not yet reached the stage of full autonomy in executing attacks against well-protected systems.

Restricted access and stringent identity requirements

Access to the Daybreak Blue and Daybreak Red levels is restricted to individuals and institutions that have undergone identity verification and legal declarations, and Open AI has announced that it will impose hardware security keys on all individual Daybreak accounts starting from September 1. Announced partnerships include SpecterOps, SentinelOne, and Palo Alto Networks, entities that work on the front lines of cyber defense, a one-clause gloss on the first mention of an institution an English reader may not recognise by its Arabic name.

What this means for the region

For the banking, government, and critical infrastructure sectors in the Gulf, the initiative raises a practical question: how can institutions benefit from models that search for vulnerabilities before attackers do, while ensuring that these capabilities do not leak to hostile parties? The partial answer lies in the governance model proposed by Open AI, identity verification, legal declarations, and security keys, a pattern that may become a reference standard for licensing dual-use artificial intelligence tools in the region.

The initiative is in its early stages, and the final number of beneficiaries has not been announced, nor has the full extent of the vulnerabilities that the model will discover. However, the signal is clear: the cyber capability race has entered a new phase, and entities that build their defenses today with the help of these tools will have an advantage when attackers have what defenders have tomorrow.