Is the strongest cyber attack model Mythos unsealed? Anthropic reveals that it will bring Mythos level models to all customers in the coming weeks According to Beating monitoring, Anthropic wrote at the end of Claude Opus 4.8's announcement, 'We also plan to release a brand new level of model with higher intelligence than Opus.'. As part of Project Glasswing, a few organizations are currently using Claude Mythos Preview for network security related work. Before this level of capability model is widely released, it needs to be equipped with stronger network security protection measures. We are rapidly advancing the development of these protective measures and expect to bring Mythos level models to all customers in the coming weeks. （We're making swift progress on developing these safeguards and expect to be able to bring Mythos-class models to all our customers in the coming weeks.） It is worth noting that there is currently a divergence in public reports regarding Anthropic's claim to "bring Mythos level models to all customers", and the specific model and definition of "all customers" still need to be discussed. On May 23rd, Anthropic reiterated in its latest software security collaboration project "Project Glasswing" that even if internal testing proves that the new generation security base model Claude Mythos Preview has extremely powerful vulnerability mining and network attack and defense capabilities, the company will indefinitely prohibit the opening of Mythos level models to the public and commercial markets until more effective security alignment mechanisms are developed. ---------------------------------Click on the original link below to join the Beating · Feishu AI news channel, which continuously monitors global AI hotspots and news 24/7. [Original link]