openai recently announced that it will open its state-of-the-art cybersecurity large model, gpt-5.5-cyber, to the european union, granting access to enterprises, government agencies, cybersecurity experts, and key entities such as the eu’s ai office. this move is seen as a crucial step for openai to strengthen global cooperation on cyber defense, leveraging the understanding and analytical capabilities of large models to help relevant organizations address increasingly complex cybersecurity challenges.
the european commission has expressed strong approval, with a spokesperson publicly praising openai’s transparency at a press conference, noting that this willingness to be open allows regulators to track model deployments in real time and more effectively manage potential risks.
in contrast, openai’s competitor anthropic has made significantly slower progress in europe. although the company launched its cybersecurity model, mythos, with great fanfare a month ago—raising concerns about software security—it still has not obtained permission to enter the eu market.
reportedly, the eu is engaged in intensive consultations with anthropic regarding access permissions, with four to five rounds of negotiations held so far. however, discussions remain at the procedural stage, with no concrete solutions yet reached. this disparity in compliance progress has resulted in markedly different trajectories for global cybersecurity large models in the eu. industry observers believe that openai’s early success in securing regulatory clearance positions it to gain a first-mover advantage in the eu cybersecurity market.
