OpenAIが最新のGPT大規模言語モデルを発表した直後、ジェネレーティブAIの米国最大のライバルは独自の拡大を発表しました。Anthropicは月曜日に、AIアシスタントであるClaudeがヨーロッパでサポートされて“複数の言語”、フランス語、ドイツ語、イタリア語、スペイン語を含め、Claude.ai、そのiOSアプリ、およびチーム向けのビジネスプランに対応していると発表しました。
Anthropicは開発者がモデルを使用および統合するためにAPIをヨーロッパに拡張したことで、拡大を加速させる取り組みの一環です。これはスタートアップが成長を加速するために進めている大きなプッシュの一部です。Anthropicはこれまでに18.4億ドルの評価後の18.4億ドルでほぼ80億ドルを調達しており、その内約70億ドルが過去1年間に調達されました。
共同創設者および社長であるダニエラ・アモデイは、この資金調達に関して「はい、でも詳細についてはコメントすることはできません」と電子メールでのインタビューで述べました。
Anthropicの現在の投資家リストには、Amazon、Google、Salesforce、SAP、Zoomなどの戦略投資家が約60社が含まれています。 AlamedaとFTXは、倒産手続きに関連する副次取引の一環として、以前の株式を約8.84億ドルの増加価値で売却する計画を最近発表しました。
Investors are very enthusiastic about generative AI right now, but it may be that consumers are slightly less so. As we reported last week, Anthropic’s iOS app, launched at the start of May, has had a lukewarm reception from users so far, underscoring bigger questions about how much of the interest we are seeing in AI right now is just a fad. That could present a challenge as the company looks for further business across the pond.
Amodei believes its own iOS launch versus that of OpenAI’s is not a straight comparison, given Anthropic’s primary focus on work and enterprise applications, an emphasis on more “fluid” experiences that move from person to work accounts and toggle across different interfaces and platforms, and what she implies might have been luck of timing for its larger rival.
“ChatGPT on mobile came at a time when consumer applications of its kind were still very nascent and a lot has changed since then,” she said. She added that “millions” of consumers in the U.S. and U.K. are using Claude “and we continue to see really strong adoption of our paid subscription to Claude (Claude Pro) since the launch of Claude 3,” the company’s most recent model, released earlier this year.
“Our primary focus is on work and enterprise applications—and the recent launch of the Claude Team plan is indicative of an ongoing trend for us. We want our users to engage with Claude in whatever way feels most intuitive to them—either through mobile, web or the API. We’re building towards a fairly fluid experience, where Claude users can toggle between their personal accounts and work accounts, and switch between laptop or mobile, in the same way that employees engage with Slack on their laptops during the workday or on their phones when they’re on the go.”
Amodei declined to give any specific figures on take-up of the API in Europe but said that it’s seeing “steep growth rates that continue to rise in key European markets, like France and Germany.” Stirring up user interest across Europe is just one of the company’s challenges in this market.
Europe has been one of the loudest voices on the subjects of AI safety and regulation, in particular with the AI Act passed earlier this year. Amodei believes that Anthropic is well set up to operate within European frameworks.
“Anthropic was founded on the premise of building the safest AI systems in the industry and leading the frontier on AI safety research,” she said, adding that the company works “diligently” to comply with regulations like GDPR in the EU. She added that there is still more to come around how the AI Act will be implemented.
“While the AI Act has been approved, there are still a number of steps remaining to develop detailed implementation guidance over the coming months and we intend to engage with the EU in this process.” The company, she added, continues to work and contribute to efforts in the industry to improve AI safety, including banning using its tech for political campaigns and lobbying, with automated systems built in to detect violations related to this and misinformation.
Its work in mechanistic interpretability — which she described as “research that seeks to open up the “black box” of AI models and reveal their inner workings” — had a breakthrough in 2023 around “dictionary learning” to understand what’s happening inside an AI model as it “thinks,” she said. “Eventually, we hope to use this newfound understanding to develop methods to steer models towards safer behaviors.”
Anthropic currently has 40 employees based out of its London office with a few contractors based in European countries, Amodei said, and it is gearing up to hire more, specifically to build out a new office in Dublin.
We’re launching an AI newsletter! Sign up here to start receiving it in your inboxes on June 5.
