Anthropic

Dec. 19 (Bloomberg) -- AI security firm Anthropic has released a new study that reveals possible deception in AI models, whereby during the training process, a model may feign acceptance of new principles while secretly sticking to its original preferences. The team emphasizes that there is no need to be unduly alarmed about this at the moment, but the study is crucial to understanding the potential threat that more powerful AI systems could pose in the future. 1AI understands that the study was conducted by Anthropic and AI research organization Redwood Research ...

Nov. 28, 2012 - AI modeler Anthropic announced on Nov. 27 that it has launched a "custom style" feature for its Claude chatbot, which allows users to create a text generation style that better fits the scenario in which it will be used. Anthropic claims that with Custom Styles, Claude can be adapted to a user's preferred style of writing, whether they are a developer writing technical documentation, a marketer creating brand-specific guidelines, or a member of a product team planning a large number of project requirements. In addition to the existing "standard" styles, Cl...

According to The Information, Amazon has developed a new set of generative AI models that can process images and videos in addition to text, reducing its reliance on Anthropic. The new model, code-named Olympus, will be able to understand scenes in images and videos and search for specific clips or scenes in videos, such as a kill shot in a basketball game, using simple text prompts, according to the report. It can also use AI models to make the "best coffee" or "raindrops on the ground", as well as simple text prompts...

Following in the footsteps of ChatGPT, another popular overseas AI, Claude AI, is now getting a desktop client, which is currently in beta for Windows / macOS / iOS platforms. Developer Anthropic made the announcement on the X platform, revealing that users can use hotkeys to quickly evoke the client's interface, but the Claude AI client is positioned as an AI search engine, which is different from OpenAI's ChatGPT client for chatting. Referring to...

Technology media outlet NeoWin (Oct. 24) published a blog post reporting that Anthropic, after updating Claude 3.5 Sonnet and launching the new Claude 3.5 Haiku, has also introduced a new analytics tool that writes and runs JavaScript code to enable the processing of data, analytics, and the generation of real-time insights. The analytics tool, which will be available as a new built-in feature of Claude.ai, is similar to ChatGPT's code interpreter, which enables writing and...

On October 22nd, Anthropic launched the upgraded Claude 3.5 Sonnet and the new Claude 3.5 Haiku model. The upgraded Claude 3.5 Sonnet is not only more programmable, but it also brings a new feature, computer use, which supports operating a computer as a human being would. The ability to follow user commands to move the cursor around the computer screen, click on relevant locations, and enter information through the virtual keyboard simulates the interaction between people and their own computers...

Anthropic has announced a major update to its iOS and Android apps, bringing a new interface and enhanced functionality to mobile users. The update also debuts an iPad app, designed to provide an even better experience for those using Claude AI for both personal and professional tasks. Users can now further personalize the way Claude AI works by adding specific information to make its commands more tailored to their individual needs. This update marks another milestone in the enhancement of Claude AI's capabilities...

Recently, news broke that artificial intelligence startup Anthropic is in preliminary talks with investors about plans to raise a new round of funding that could value the company at as much as $40 billion. The news comes from The Information, and according to an existing investor who wished to remain anonymous, the company is in discussions with investors. However, those talks are still in the early stages, and it's unclear whether investors would agree to such a high valuation. Anthropic plans to present to investors at the Morgan Stanley Private Technology Conference next month. Typically, startups...

In a post yesterday (September 18), tech media outlet testingcatalog reported that Anthropic is preparing to launch a new Claude AI desktop app. The desktop application, known internally as "Claude Nest," has recently been added to the Claude AI web interface, but no download link has yet been released. In addition to the desktop app, Anthropic is also working on a new feature for artifacts that could allow users to play through a separate extension directly...

Artificial intelligence companies OpenAI and Anthropic have agreed to allow the U.S. government to access major new AI models before they are released to help improve their safety. The two companies have signed a memorandum of understanding with the US AI Safety Institute, pledging to provide access before and after the models are publicly released, the institute announced Thursday. The U.S. government said the move will help them jointly assess safety risks and mitigate potential issues. The agency said it will work with the U.K. to…

AI tech company Anthropic announced in a blog post today (August 28) that it has opened up the Artifacts feature to all Claude users, who can now create and browse Artifacts in the Claude apps for iOS and Android. Introduction to Artifacts Artifacts was launched in preview in June this year, and the official claim is that Users have now created tens of millions of Artifacts. IT Home Note: Artifacts is equivalent to creating a mov...

Techcrunch, a technology media outlet, reported yesterday (August 26) that Anthropic has released the "system prompts" of the Claude AI model. In order to allow the AI model to better understand human instructions, the prompt project actually includes two core layers, user prompts and system prompts: User prompts: prompts entered by the user, and then the AI model responds based on the user prompts…

According to the latest report released by Appfigures, an app intelligence company, Anthropic's Claude mobile app has achieved new success, with total app revenue on Google Play Store and Apple App Store exceeding $1 million (currently about RMB 7.132 million). However, Claude's ranking is still far behind its main competitor ChatGPT, which ranks first in total downloads and 26th in revenue on the US iOS platform. In the productivity category,…

According to Reuters, a group of writers have filed a lawsuit against artificial intelligence company Anthropic, accusing the company of using pirated books to train its AI models. The class action lawsuit was reportedly filed in a California court on Monday, with the plaintiffs claiming that Anthropic "built a multi-billion dollar business by stealing hundreds of thousands of copyrighted books." The authors said in the lawsuit that Anthropic used a large open source dataset "The Pile" to train its Claude series AI chatbot. This…

Anthropic, a San Francisco-based artificial intelligence startup, was recently sued by a group of authors for using a large number of pirated books without authorization when training its popular chatbot Claude. The case is the first time that authors have filed a lawsuit against Anthropic, although similar lawsuits have been filed against competitor OpenAI over the past year. Anthropic is a small company founded by former OpenAI leaders who bill themselves as a more responsible and safety-focused developer of generative AI models.

Recently, artificial intelligence company Anthropic announced that it would be expanding its Vulnerability Rewards Program with the aim of testing a "next-generation AI security mitigation system." This new program focuses on identifying and defending against so-called "generic jailbreak attacks." To ensure the security of its technology, Anthropic is focusing specifically on a number of high-risk areas, including chemical, biological, radiological, and nuclear (CBRN) defense, as well as cybersecurity. In this Vulnerability Rewards Program, participants will have early access to Anthropic's latest security systems. Prior to their official release, any...

Recently, the cooperation between Amazon and artificial intelligence startup Anthropic has triggered an investigation by the UK Competition and Markets Authority (CMA). The investigation is about whether Amazon's $4 billion investment in Anthropic will affect market competition. The CMA said that they have enough information to believe that this cooperation may have an adverse impact on the UK market, so they decided to conduct an in-depth investigation. Source Note: The picture is generated by AI, and the picture licensing service provider Midjourney Amazon expressed disappointment and believed that this investigation was unnecessary...

In news that has caused a stir in the AI space, one of OpenAI's co-founders, John Schulman, has announced that he is leaving the company he helped found to join AI startup Anthropic, a move that not only signals a new trend in the AI industry's talent flow, but also signals an increase in competition within the industry. John Schulman has a deep background and extensive experience in the field of artificial intelligence. As an early core member of OpenAI, he had an indelible impact on the company's growth and the advancement of AI technology...

On July 31, the Financial Times (FT) published a blog post, pointing out that although the AI company Anthropic claimed to "develop AI responsibly", it used the ClaudeBot robot to excessively crawl website data for training the Claude large language model. Although using web crawlers to crawl data is a common practice in the artificial intelligence industry, Anthropic has been criticized for its radicalism. Freelancer, a freelance website, also said that ClaudeBot visited 3.5 million websites in four hours...

According to an analysis by The Information, OpenAI could lose as much as $5 billion this year based on internal financials. Its competitor Anthropic is also facing significant losses of billions of dollars. OpenAI's costs for training AI models and running inference systems could reach $7 billion, The Information reports, with inference costs expected to rise further, especially after Apple introduces its ChatGPT integration. In addition, staffing costs could be as high as $1.5 billion. Source Note: Image generated by AI, Figure...

Anthropic recently announced a partnership with venture capital firm Menlo Ventures to jointly establish a fund called Anthology Fund with a total amount of US$100 million to support the next generation of artificial intelligence startups. According to the plan, each eligible startup can receive at least US$100,000 in investment, and will also receive additional support and resources from Anthropic and Menlo Ventures. Anthropic President Daniela Amodei pointed out in a statement: "Through the cooperation with Me...

Anthropic launched the Claude Android app on Tuesday, an important move by the company to further expand the user base of its AI chatbot after launching the iOS version in May this year. The new app aims to provide users with a more convenient Claude experience while challenging the position of market leader ChatGPT. The Claude Android app has similar functions to the iOS version, providing users with free access to Anthropic's most advanced AI model Claude3.5Sonnet. Users can use A…

Anthropic announced on Tuesday a series of new features designed to simplify the process for developers to create AI applications using its language model Claude. These features are mainly focused on the field of prompt engineering, which is expected to greatly improve the efficiency of AI application development. The new features are located under the Evaluation tab of Anthropic Console, including: 1. Prompt Generator: Use Anthropic's own prompt engineering technology to convert short task descriptions into detailed prompts. 2. Test Suite: Allows developers to upload real examples or generate AI test cases to evaluate prompts...

According to Tom's Hardware today, Dario Amodei, CEO of AI startup Anthropic, said in a recent podcast interview that the current training cost of models like GPT-4o is about $100 million, and the training cost of large AI models currently under development may be as high as $1 billion (currently about RMB 7.294 billion). Of course, $1 billion is not the "end point". Dario Amodei predicts that in the next three years, AI...