MoonPay’s PayBox vault lets ChatGPT and Claude users authorize crypto purchases, token swaps and crosschain transactions while retaining custody of their assets. MoonPay has launched PayBox, a payment vault that lets ChatGPT and Claude users authorize AI to buy crypto, swap tokens and bridge assets from within a conversation while keeping control of their funds. Users connect a crypto wallet and payment methods to ChatGPT or Claude, which prepare transactions such as token swaps, crosschain transfers and DeFi interactions using natural-language prompts. Users can approve each transaction with a passkey or set spending limits that allow the AI to execute certain actions automatically, according to MoonPay. MoonPay said PayBox uses multi-party computation and trusted execution environments to protect wallet keys, preventing either the AI assistant or MoonPay from independently accessing user funds. The platform also lets users require approval for every transaction or allow AI to act autonomously within predefi...
Co-Invest lets users fund accounts, analyze positions and execute trades across more than 500 markets without leaving AI chat interfaces. Liquid, a multi-asset trading platform, has launched a trading app that lets users execute trades directly inside OpenAI ChatGPT and Anthropic Claude across crypto, equities, foreign exchange markets and prediction markets. According to the company, the Co-Invest app's users can fund accounts, analyze positions and place trades without leaving the chat interface. Liquid said the platform routes orders through venues including Hyperliquid, Lighter and Ostium. Liquid said Co-Invest supports trading across more than 500 markets, including pre-IPO secondaries and positions on Polymarket. The company said that its platform has processed more than $3 billion in trading volume since launching in August 2025 and currently serves roughly 40,000 users. Read more
In an experiment, a chatbot resorted to blackmail after it found an email about replacing it, while in another, it cheated to complete a task with a tight deadline. Artificial intelligence company Anthropic has revealed that during experiments, one of its Claude chatbot models could be pressured to deceive, cheat and resort to blackmail, behaviors it appears to have absorbed during training. Chatbots are typically trained on large data sets of textbooks, websites and articles and are later refined by human trainers who rate responses and guide the model. Anthropic’s interpretability team said in a report published Thursday that it examined the internal mechanisms of Claude Sonnet 4.5 and found the model had developed “human-like characteristics” in how it would react to certain situations. Read more