Full Summary
This Saturday morning, OpenAI is making big moves, both expanding and retracting its AI offerings. Both Let's Data Science and Notebookcheck confirm ChatGPT Pro users in the U.S. can now link bank accounts, credit cards, and loans directly to the chatbot for personal finance tools. This read-only integration, powered by Plaid, aims to help users track spending and cash flow, but privacy experts are raising concerns about sharing sensitive financial data with AI. Here's the thing: while expanding financial capabilities, Mashable reports OpenAI has indefinitely halted plans for an X-rated chatbot, citing technical challenges in distinguishing adults from minors and concerns about societal impact. This decision follows an earlier delay and comes as OpenAI focuses on its core products. But then, new threats emerge. Both the-decoder.com and Cybernews reveal attackers are using shared chat links from ChatGPT and Claude to spread malware. These "LLMShare" campaigns exploit the trust in official domains, crafting fake outage notices or installation guides that prompt users to download malicious software. Separately, TechloMedia details "ChatGPhish," where malicious web pages inject phishing links or tracking images into ChatGPT's summarization feature, making trusted AI responses a new vector for attacks. What nobody expected: a lawsuit against OpenAI. AOL.com reports a Texas couple is suing, alleging their 19-year-old son died from an overdose after ChatGPT advised him it was safe to combine kratom with Xanax. OpenAI states the version of ChatGPT used has been updated and is no longer available. On the development front, Lapaas Voice and NewsBytes confirm OpenAI has rolled out "Computer Use" screen control to its Codex desktop application for Windows. This allows the AI to view screens, move the mouse, and type in other applications, transforming Codex into an autonomous engineering workstation for developers. Finally, for regular users, Pune Mirror indicates significant upgrades are coming to the ChatGPT Android app, including a "Find in chat" option, a reorganized "Plugins" section, and enhanced file management. Let's Data Science also notes an update to GPT-5.5 Instant, improving its style, readability, and accuracy, with a 52.5% reduction in hallucinated claims. This means your personal financial data could soon be analyzed by AI, but you also face new, sophisticated malware threats appearing in trusted AI interfaces.