Anthropic is preparing to broaden access to its Claude Mythos AI model, which was previously available only to select companies such as Amazon, Microsoft, and Apple under Project Glasswing. Following the launch of Claude Opus 4.8, the company plans to offer Mythos to all customers. According to prediction markets on the Kalshi platform, over $390,000 has been wagered on the timing of Mythos's release, with a significant probability of it being available before August 1, 2026. Bettors are particularly optimistic about an earlier launch, with a 71% chance of it being released before July 1, 2026, and a 39% chance before June 15, 2026. Cloudflare, a participant in Project Glasswing, has commended Mythos for its ability to construct exploit chains but raised concerns about its safety due to inconsistencies in processing certain requests. Companies like Cisco and Palo Alto Networks are also expected to leverage the advantages of Mythos AI. Anthropic's emphasis on recursive self-improvement in AI highlights the ongoing evolution of this technology.