Anthropy: Claude 2 is here

Anthropic, an artificial intelligence startup, has released a new version of its AI robot called Claude. Claude 2 is expected to bring improved performance and longer responses, and will be accessible via an API and a new public beta site, claude.ai. However, this currently only works in the US and UK.

Claude 2: Longer answers, more security

One says, “Compared to our previous models, we’ve improved in coding, math, and logical reasoning.” Blog entry by Anthropic. Like its predecessor, Claude 2 is based on a rich language model and can be used for written tasks such as abstracts, research, question answering, and coding. Both models are currently said to be able to capture “large chunks of text”, but Claude 2 is said to be able to generate longer responses than its predecessor. According to the company, responses can be up to 3,000 words long.

Claude 2 will also provide more specific answers on some topics, such as coding and elementary-level math, according to Anthropic. In addition, the team has set itself the goal of making Claude “less vulnerable to manipulation” than other chatbots: “We’ve always worked to improve the security of Claude 2, so that it’s more malicious and harder to use abusively or generate serious expense,” says Anthropic.

Anthropic and the law of artificial intelligence

Think of Claude as “a friendly, enthusiastic colleague or personal assistant” who can be taught in natural language to “help out with many tasks,” the blog continues. However, this currently only applies to users in the USA and UK: “Cloud 2 […] It is generally available in the United States and the United Kingdom. We are working on making Claude available globally in the coming months.”

There is no further explanation as to why Claude is not yet available in this country. EU artificial intelligence law could play a role. It is trying to walk the fine line of allowing innovation with AI on the one hand, but at the same time dominating powerful technology. There will be strict rules and requirements, especially for the so-called foundation models, on the basis of which applications work – such as Cloud. Claude gave a poor score of 7 out of a possible 48 points in evaluating currently available models for compliance with AI law (tell us).

Just this past May, Anthropic raised a whopping $450 million in a Series C round. In addition to Sound Ventures, tech giants like Google, Salesforce, and Zoom also participated in the funding (more info here).

New Google Privacy Policy: In the future, AI will be allowed to learn through our content