OpenAI has announced Prism, a new AI-driven product developed specifically for academic work. With Prism, OpenAI does not appear to be focused only on improving academic writing. Instead, the company seems to be aiming to reclaim market share it has started to lose following the growing popularity of learning and research-oriented tools such as NotebookLM and Anara in recent months. January 2026 data showing a decline in traffic referred by ChatGPT, alongside an increase in traffic coming from Google’s Gemini has made this competitive shift more visible.
Prism describes itself as “a collaborative LaTeX editor” and brings academic writing, source & citation management and scientific visualization features together within a single platform. Rather than resembling a conventional AI assistant, the product has an interface specifically designed for researchers and academic workflows.
Within just a few days, Clawdbot (now renamed Moltbot) gained rapid traction, especially across developer communities on social media, with its Github Repo approaching 80K stars in a short time. Positioned as an open-source alternative to Claude Code and Claude Cowork, Moltbot has attracted developers because it functions as an assistant not only for coding, but also for other digital workflows. In a space increasingly dominated by closed ecosystems, being open source is a key part of its appeal.
As usage and visibility surged over the past few days, the assistant quickly underwent a rebrand following a trademark request related to Claude by Anthropic. Clawdbot announced via a post on X that the project would continue under the new name Moltbot.
Elon Musk’s so called “source of truth,” Grokipedia, has now referred as a source by ChatGPT. According to tests conducted by The Guardian, ChatGPT tends to rely on Grokipedia particularly when responding to niche or lesser-known topics. The platform is not considered a reliable source and it is criticized since its launch for providing biased information and allegedly copying large portions of Wikipedia’s database.
The risks posed by AI systems referencing one another are significant for a reason: In this case, it means that a model (GPT-5.2) is treating AI-generated content as a primary source. That output can then be mixed into other models’ responses and eventually recirculate across the web. Over time, misinformation may start to appear credible simply because it has been repeated often enough. 🤖
Anthropic announced that it has integrated some business applications directly into Claude. With this update, users can now access tools such as Figma, Canva, Asana, Amplitude and Slack directly within the Claude chat interface, similar to how integrations work in ChatGPT.
In parallel, the company also released Claude for Excel, a move that appears to achieve what Microsoft Copilot has struggled to deliver for some time. As a result, simply being “good at Excel” may no longer be a sufficient qualification in the near future.
