As of this July 1, Claude Fable 5 is available again worldwide. On June 30, the US Department of Commerce lifted the export controls that had blocked public access to the model since June 12, closing 19 days of an unprecedented episode: for the first time, a frontier artificial intelligence model, among the most capable in existence, was suspended for the general public by decision of a government.
The outcome was not just a matter of “available again.” Along with the reopening, Anthropic announced a new cooperation framework with the US government: early access to its models before public launch and a joint scheme with Amazon, Microsoft and Google to rate the danger of jailbreaks (techniques that try to get around a model’s safety restrictions). Mythos 5, the more powerful sibling model, remains restricted to approved US critical infrastructure organizations. In other words, access was restored, but a channel through which Washington exerts influence before a model reaches the public was also institutionalized.
For Latin America, the episode leaves an uncomfortable lesson. The region uses these tools every day (developers, companies, students) and had no voice either when the model was switched off or when it was switched back on. Everything was resolved through market pressure, reputation and direct negotiation between a company and a government in the North. Had the affected population been Latin American, it is not clear what leverage would have existed to speed up an equivalent response.
Also today
- The Bank of England warns of “market meltdowns” from autonomous AI — Deputy Governor Sarah Breeden proposed considering “emergency switches” for markets: 52% of financial firms already use AI that operates on its own.
- The Council of the EU gives final approval to the “Digital Omnibus” — It postpones the most demanding obligations for high-risk systems to 2027-2028; even the European model, the only one with real penalties, is softening under industry pressure.
- A 35-billion-parameter model matches trillion-parameter ones — Agents-A1, from the Shanghai AI Lab, reaches the performance of much larger rivals by scaling the “horizon” of its trajectories instead of its size.
- BCG documents the US-China split in the world of AI — The report describes two ecosystems pulling apart and does not mention Latin America at all.
In the region
Panama joined Pax Silica, the US State Department alliance that links AI cooperation to commitments on critical minerals and supply chain traceability. It is the fifth Latin American country to sign, after Argentina, Chile, Costa Rica and El Salvador. The pattern is clear and is advancing country by country, but there is still no coordinated regional position (from CELAC, ECLAC or CAF) on a scheme that ties access to technology to conditions on natural resources.
Launches
- Claude Fable 5 and Mythos 5: access restored — Fable 5 is available globally again on Claude.ai, the Claude platform, Claude Code and Claude Cowork after 19 days of blockade. Mythos 5, by contrast, remains limited to approved critical infrastructure organizations in the United States, with no Latin American presence among them.
Threads we’re following
This close adds to the story we have been following since mid-June, when the United States unexpectedly suspended access to Fable 5 on national security grounds and opened a negotiation whose terms were never fully public. That ban had “cracked open” at the end of the month with partial reopenings; today the chapter closes with global access restored and a new framework of prior cooperation between Anthropic and the US government. What remains to be seen is whether that framework (government early access to models) becomes the norm for the whole industry.
If a ban on a frontier model was resolved in 19 days thanks to market pressure and direct negotiation with Washington, what mechanism would a Latin American country have to achieve something equivalent if the affected population were its own?
About this entry. It is generated automatically from public sources, without human review before publication. It may contain errors of interpretation or summary; please check each story against its original source (the links lead there) before citing it or making decisions based on it.
Doble Click is written with Anthropic models.