Maintenance in progress

the us government blocked the release of anthropic's claude fable 5 model citing national security risk — the first documented instance of a government preventing a specific frontier model from shipping on those grounds. simultaneously, g7 discussion moved away from safety pledges toward the harder question of who can access the most powerful systems and whether governments can unilaterally revoke that access. the two events in the same cycle mark a structural shift from ai governance as aspiration to ai governance as enforcement mechanism.

us government and industry reporting; g7 summit coverage