Anthropic today launched Claude Fable 5, the first fifth-generation Claude model available to the general public, immediately accessible worldwide. Based on the Mythos architecture (previously reserved for institutional and government customers) but with guardrails adjusted for broad use, its arrival was confirmed by journalist Alex Heath at The Verge and in prediction markets (94.4% probability) before Anthropic published a formal announcement. As of this edition’s close, the company had not published a system card or official comparative benchmarks.
The context matters: Anthropic filed a confidential S-1 with the SEC on June 1, which indicates that it is in the process of going public. Every product announcement in this period is also, inevitably, a communication aimed at investors. The commitments about which capabilities of the Mythos model were “reduced” in Fable 5 are, for now, verifiable only by the company itself. For Latin America, this means immediate access to the model with the greatest offensive potential among the so-called frontier models without the region having taken part in Project Glasswing, the program through which Anthropic distributed cyber defense countermeasures to 78 countries before this launch. The “Code with Claude Tokyo” event, scheduled for June 10, could include the first official demonstration.
Also today
-
Colombia: new deepfake documented 12 days before the runoff — The Foreign Ministry rejected foreign interference, but Law 2502 regulating AI in campaigns still has no operational enforcement until July, after the June 21 election. It is the fourth documented incident in this campaign. (Source: Colombian media.)
-
SpaceX SPCX: IPO expected on June 11 at $135 per share — The S-1 filed with the SEC revealed that SpaceX’s current contract with Anthropic amounts to $1.25 billion a month. Estimated valuation: $1.75 trillion. Trading on Nasdaq starting June 12. (Source: SpaceX S-1 filed with the SEC.)
-
GPT-5.6 spotted in Codex logs — Several users posted evidence of a new model identifier in OpenAI’s API; prediction markets put it at 89% probability, but as of this edition’s close there are only secondary sources and no official announcement. It is an early signal, not a confirmed fact.
-
Perplexity raises an additional $200 million for Comet — The round brings the company’s valuation to $20 billion and its total capital raised to $1.72 billion. Comet is Perplexity’s browser with built-in AI; its version with local-cloud inference is planned for July 2026. (Source: trade media.)
In the region
This month Argentina became the first country in the world to grant legal personhood to companies run entirely by AI agents, without yet having a general AI law in force. The legal paradox is concrete: an “automated company” in Argentina today has more clearly defined legal certainty than any person who interacts with it. The debate between President Milei and writer Yuval Noah Harari, published in the Financial Times, positioned the country as a voluntary laboratory for AI deregulation, at a time when the rest of the region is watching to see which model to adopt for its own frameworks. The unanswered question is what happens when an “automated company” causes harm and its liability is limited to the entity’s own assets: there is not yet a clear protection mechanism for affected Latin American citizens.
Launches
-
Claude Fable 5 (Anthropic) — The first general-purpose fifth-generation Claude model. Mythos architecture with adjusted guardrails. Immediate global access; available to developers from the day of the announcement. No system card or public benchmarks at the time of launch.
-
PhysicsX Large Physics Models (PhysicsX) — Large-scale models for physics simulation, aimed at engineering, manufacturing and industrial design. Available via API. The company closed a $300 million Series C round led by Temasek, with participation from Nvidia, Siemens and Applied Materials.
-
Comet (Perplexity) — A browser with built-in AI for hybrid inference (local + cloud). Available globally; the version with full local inference is expected in July 2026.
Threads we’re following
Colombia has for weeks been the region’s most concrete case at the intersection of generative AI and electoral democracy. Since late May we have documented that Law 2502 was approved without operational enforcement mechanisms for the pre-election period. Today that scenario repeats itself: another fake image circulates, the Foreign Ministry rejects foreign interference, and the law still has no capacity to act. The June 21 runoff will be the real test of whether it is possible to defend a Latin American presidential election from generative AI without effective regulation in place.
Anthropic alone decides which capabilities are “reduced” in Fable 5 relative to Mythos, with no external evaluator to verify it and no system card to document it. Is that artificial intelligence governance, or is it governance theater?
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.