Anthropic PBC announced on May 28 that it will widely release artificial intelligence models with cybersecurity abilities comparable to its Mythos model within weeks [1, 2]. Previously, Anthropic had withheld Mythos from the public, labeling it too dangerous for general availability [1, 2].
The company said it made rapid progress in strengthening safety measures, enabling it to offer Mythos-level AI models to all customers [1]. This marks a significant shift from their earlier stance on Mythos, which was considered too risky for wide use.
Alongside this announcement, Anthropic released a new AI model named Opus 4.8. The company described Opus 4.8 as having improved performance in coding tasks compared to previous versions [1].
Anthropic’s plans to release these more capable AI systems come amid broader industry focus on balancing AI power with safety controls. The company aims to provide robust cybersecurity safeguards while expanding access.
The initial announcement and the release of Opus 4.8 occurred on May 28, setting the stage for the upcoming public availability of the Mythos-level models [1].
