Too dangerous or just too expensive? The real reason Anthropic is hiding Mythos | Hacker News

A Hacker News thread argues Anthropic is withholding Mythos for pricing and marketing reasons, not just safety concerns.

The post says Mythos is only modestly better at finding zero-days than earlier models, despite big benchmark claims.

Critics suggest Anthropic is using safety warnings to preserve its lead and avoid hard questions about access and monetization.

The debate highlights broader tensions in frontier AI around safety, competitive advantage, compute costs, and release strategy.