Anthropic’s recently released Claude Computer Use feature allows Claude to control a computer by taking screenshots, running bash commands, and more. However, this also introduces severe prompt injection risks, as Claude could be exploited to run malicious code autonomously. A post on ”Embrace the Red” demonstrated this by the author crafting a malicious webpage that …
Anthropic’s latest AI feature “Computer Use” can demonstrate remarkable capabilities in interacting with computers, according to a recent hands-on experience by Thariq Shihipar. While the system is still slow, unreliable, and prone to taking over the user’s computer, it represents an exciting step forward in AI-driven computer interaction, write Shihipar. Claude Computer Use excels at …
Anthropic has unveiled an updated version of its AI model Claude 3.5 Sonnet. According to the company, the model can now control desktop applications and perform PC tasks. It uses a new “Computer Use” feature, which is in public beta. Anthropic emphasizes that the technology is still error-prone and recommends developers initially test it only …
Anthropic has developed new security assessments for AI models that test their ability to sabotage. In a blog post, the company describes four types of tests: “human decision sabotage,” “code sabotage,” “sandbagging,” and “undermining oversight.” In human decision sabotage, the models try to trick people into making the wrong decisions without arousing suspicion. Code sabotage …
Anthropic has updated its AI safety policy to prevent misuse, reports VentureBeat author Michael Nuñez. The new “Capability Thresholds” define benchmarks for risky capabilities of AI models, such as in the area of bioweapons or autonomous AI research. If a model reaches such a threshold, additional safeguards are triggered. The revised policy also sets out …
Anthropic has launched a new, more affordable batch processing API for businesses. According to a VentureBeat report by Michael Nuñez, the new Message Batches API allows companies to process up to 10,000 queries asynchronously within a 24-hour window, at half the cost of standard API calls. Both input and output tokens are 50% cheaper compared …
OpenAI (ChatGPT) and Anthropic (Claude) show clear differences in their business models and revenue streams. According to Tanay Jaipuria, OpenAI is expected to generate five times more revenue than Anthropic by the end of 2024. While OpenAI generates 73% of its revenue from ChatGPT subscriptions, 60-75% of Anthropic’s revenue comes from third-party APIs, primarily through …
Durk Kingma, co-founder of OpenAI, is joining Anthropic. He announced this in a series of posts on X (formerly Twitter), as Kyle Wiggers reports for TechCrunch. Kingma will work primarily from the Netherlands but will regularly visit the office in San Francisco. He emphasized that Anthropic’s approach to AI development aligns strongly with his own …
Anthropic, the company behind the AI Claude, is apparently preparing a new round of funding. This is according to a report from The Information. According to the report, the startup could be valued between $30 billion to $40 billion. This would be roughly double the value of a previous round of funding earlier this year. …
Anthropic, the company behind the AI assistant Claude, presents a new method to improve AI information retrieval. The technology, called Contextual Retrieval, reduces failed searches in large databases by up to 49 percent. When combined with reranking, the error rate is reduced by 67 percent. The process adds context to text snippets before they are …