The chatbot, named Claude, will compete with OpenAI’s ChatGPT.
(Bloomberg) — Anthropic, an artificial intelligence startup positioning itself as the builder of a safer kind of chatbot, said it raised $450 million in funding to bolster the development of its AI bot, named Claude. The deal brings the company’s total raised to more than $1 billion.
Claude, like OpenAI’s ChatGPT, is built atop a large language model and can be used for written tasks like summarizing, searching, answering questions and coding. Yet while ChatGPT has faced criticism — and been tweaked — after offering users some concerning results, Anthropic aims to make its chatbot more cautious from the start.
Anthropic’s goal is for Claude to be less prone than other chatbots to manipulation that will produce harmful or offensive results. The startup’s co-founders are former leaders from competitor OpenAI.
Spark Capital led the funding round, which also included Alphabet Inc.’s Google, Salesforce Ventures, Sound Ventures and Zoom Ventures, the company said. The investment follows a nearly $400 million investment from Google that Bloomberg reported in February. The company declined to discuss its current valuation.
Anthropic was founded in 2021 by a handful of former OpenAI staffers, including siblings Daniela Amodei and Dario Amodei. The startup and its leadership have urged caution about the rush to develop and release AI systems, and their potential to transform industries.
“There’s incredible potential that I think comes from these systems, but also, as we’re all seeing, there’s really big potential for harm,” Daniela Amodei said. “I think we don’t yet completely understand all of the potential for negative externalities that these systems can create. We’re really just starting to see some of them.”
Anthropic began offering its chatbot — its first publicly available product — to businesses that want to add it to their products in March. Claude is currently available through apps such as Quora Inc.’s Poe, which lets users ask questions and discuss them. It’s also on offer through Slack, via an app that can provide summaries of Slack threads and answer questions.
Anthropic plans to use the fresh cash influx for continued training of Claude and future AI models, Daniela Amodei said, and to bring its chatbot to more business users. Anthropic announced a partnership with Google in February through which it’s using Google’s cloud computing service to train and roll out its AI models. The company’s cloud partners are Google and Amazon Web Services, according to a spokesperson.
“It’s very capital-intensive to train frontier AI models, and that’s generally why you’ll see the majority of these large systems being trained at post-IPO, publicly traded, well-established companies,” Daniela Amodei said.
Although chatbots have been around for a while, Claude is part of a wave of newly powerful tools that have been trained on massive swaths of the internet to generate text that mimics human speech far better than predecessors. Such tools are an application of generative AI, which refers to artificial intelligence systems that consider input such as a text prompt and use it to output new content such as text or images.
OpenAI released ChatGPT for widespread use in November, and its sudden popularity spurred numerous tech companies to push out chatbots, too; these include Google’s Bard and OpenAI investor Microsoft Corp.’s Bing chat, which uses OpenAI’s technology.
Daniela Amodei said Anthropic plans to increase its investments in AI safety research with the new funding. “What we have found is being able to stay at the frontier in terms of developing more capable models often actually illuminates new safety challenges that we didn’t see at smaller scale,” she said.
Yasmin Razavi, a general partner at Spark Capital, is also joining Anthropic’s board, the company said, which increases the size of its board to four members.
Razavi said part of her interest in the company stems from its safety-related efforts. She also sees Anthropic as building a foundational technology at a time when the AI field is evolving rapidly.
“Being at that foundational layer today, at least from my perspective, gives you a chance to see a second act and a third act and a fourth act of the industry,” she said, “and it’s moving incredibly quickly.”
More stories like this are available on bloomberg.com
©2023 Bloomberg L.P.