Claude is now available in Europe

Original Article ↗ Hacker News Discussion ↗

Claude AI’s expansion to the EU prompts mixed reactions: many developers praise its coding and refactoring abilities, often finding it stronger than GPT‑4 in complex software tasks, while others complain about aggressive web crawling, strict trust‑and‑safety bans, and friction in account setup such as mandatory phone verification. Commenters also debate broader issues of AI moderation and “censorship,” compare proprietary models to open-source alternatives like Llama 3, and note regional gaps and quirks in availability, including prior unofficial access in parts of Europe and continued absence in Canada.

Web crawling and bot behavior

  • Some report Claude’s crawler as among the most aggressive on their servers; one mentions 60 URLs/s across ~100M pages, more than they see from Googlebot.
  • Others argue this rate is actually low relative to total pages and not close to a DDoS.
  • Several compare Claude’s bot with Google, Bing, Semrush, etc.; Google is noted as far more frequent but also as a source of traffic, whereas Claude is seen as “only taking.”
  • Complaints also arise about other bots (e.g., Semrush) ignoring robots.txt and being more abusive than Claude.

Account bans, trust & safety, and “censorship”

  • A user reports being permanently banned after asking for a translation of Japanese text on a kitchen knife and getting no reply from support.
  • Others describe similar inexplicable bans that were later reversed.
  • This triggers a broad debate about trust & safety teams, perceived overreach, and whether corporate moderation is “censorship,” “fascism,” or necessary risk management.
  • There is extended discussion about culture-war issues (DEI, political correctness), free speech, and whether LLM safety policies suppress diversity of viewpoints.

Voice interaction and UX

  • Some see GPT’s two-way voice interface as a “killer feature,” hoping Claude will at least offer good speech-to-text.
  • Others find conversational voice unnatural, time-pressuring, and even unsettling, but acknowledge its accessibility potential.

Model quality and coding performance

  • Multiple commenters say Claude outperforms GPT-4/4o for their software development tasks, especially complex refactoring and working through obfuscated JavaScript.
  • A cited leaderboard suggests Claude leads GPT-4o on harder refactoring tasks, but a subthread questions how much benchmarks are distorted by training data leakage and overfitting.

Access, geography, and sign-up friction

  • Several Europeans report having used Claude earlier despite official non-availability; one notes occasional “not available in Europe” popups that didn’t block use.
  • Clarification that the launch is specifically for the EU; the UK already had access.
  • A UK user reports SMS verification failures.
  • Some dislike that Claude.ai requires accounts and mobile numbers, viewing this as user-hostile and steering them toward open models. Workarounds like cheap prepaid SIMs are mentioned.
  • Confusion over Anthropic’s statement that the API is “not intended for individual use” leads some to stick with OpenAI or open-source APIs.
  • Commenters from Canada note Claude still isn’t available there and question why it’s accessible in the EU but not Canada.