Anthropic's Glasswing ushers in gated frontier AI

This feels less like a consumer product launch and more like a controlled cyber-defense program, so the gatekeeping is intentional rather than accidental.

• –The narrow partner list is real, and it includes major incumbents, which makes the poster’s concentration concern reasonable.
• –Anthropic is explicitly framing this as dual-use risk management, not a public launch with open signup.
• –The trend is not new in AI or tech generally: the most capable systems often roll out first through invite-only or enterprise-only channels.
• –What is new is how clearly frontier capability and security posture are now linked, with access shaped by perceived misuse risk as much as by price.
• –The anti-competitive read is plausible as an interpretation, but the public evidence here supports security gating more directly than deliberate exclusion of smaller rivals.