Slovenian domain demand surges amid Trump’s AI rebrand

Wider adoption remains a bet: the naming order covers federal agencies, not businesses, and registration totals conflict.

By Saeed EzzatiPublished Oct 4, 2026, 2:00 AM PDT8 min read

Slovenian domain demand surges amid Trump’s AI rebrand

The audio edition

About 3:00

0:003:00

Read transcript

Slovenia’s .si web addresses are suddenly in demand, just as President Trump is urging US federal agencies to replace “AI” with “SI,” or “Super Intelligence.” Register.si told the BBC it recorded 44,000 registrations in September, up from fewer than 2,000 in August, including 11,000 on September 30. TechRadar cited a different count: 46,066 in September versus 3,515 in August. Both accounts show a steep rise, but the exact total is unresolved. The timing has fueled a link to Trump’s rebrand, and the .si suffix matches his preferred initials. But the registry has not established that his comments caused the rush. The federal directive applies to executive-branch communications outside statute; it does not require companies or the wider AI industry to adopt the term. One possible explanation is domain speculation: buyers reserve names hoping to resell them if demand grows. That is not automatically cybersquatting. TechRadar says the registry charges registrars ten euros per domain, while Register.si dismissed claims of millions in Slovenian earnings as having no real basis. Also, TechRadar’s “over one hundred times” headline compares a busy day with an average August day—not monthly totals. For domain buyers, wider business adoption remains a bet, not a result guaranteed by government wording. That gap between a striking signal and a real-world result also appears in accounting. Mercor says frontier models beat 12 licensed accountants on speed and accuracy, scoring perfectly on four simplified month-end tasks; models cost more than ten times less per grading criterion met. But the test excluded client conversations, coworker help and company context, and did not measure human-AI collaboration. The broader APEX-Accounting benchmark remains difficult: The Decoder reported almost 60 percent of tasks had not been fully solved by any model. And turning model performance into useful work takes people as well as software. Anthropic is committing $100 million to its Claude Frontier Academy, aiming to train 10,000 engineers by the end of 2027. Organizations nominate experienced engineers, who must pass a practical assessment before leading a real Claude project during a 12-week residency. A second assessment determines the final credential, with the first expected in early 2027. The target is a training goal, not a count of qualified graduates so far. That question of what works beyond a controlled test matters especially when AI looks human. In Tavus’s company study, 26 of 54 participants mistook Griffin-Lite for a person after a one-minute video call. They had expected to meet another human; the small test does not show whether the impression lasts in longer calls or whether the system provides useful assistance. Griffin-Lite remains a research preview for trusted testers while Tavus develops safety and disclosure features. There is no firm public launch date, and identity-copying consent checks remain unresolved. Across these stories, watch what people and businesses actually adopt—and what holds up outside a narrow test.