It’s official: 900 million people are now flocking to ChatGPT each week for AI-powered writing, answers, thinking and more.
Most of those people use the free version of ChatGPT, while about 50 million users access the AI via a paid subscription, according to writer Aisha Malik.
Adds Malik: “The new weekly active user figure marks a jump of 100 million users from the 800 million that OpenAI reported in October 2025.”
In other news and analysis in AI writing:
*Meet Your Even Tougher Workplace Competitor: ChatGPT-5.4: OpenAI says its newest AI engine – GPT-5.4 — matches or outperforms many human professionals 83% of the time, according to the maker.
The new model is also more accurate and is 18% less likely to generate errors and 33% less likely to come back with false claims, according to OpenAI.
GPT-5.4 is currently rolling out as ChatGPT-5.4 and is also available via direct computer API access.
*U.S. Military, Government Pivots Away from Anthropic Products: Angry that AI titan Anthropic refused to share full access to its tools for any legal purpose, President Donald Trump has banned use of Anthropic products by the U.S. government.
Observes lead writer Kali Hays: “The company had grown concerned in recent months about the government potentially using its AI tools — like Claude — in what it described as mass surveillance and fully autonomous weapons.”
Claude and similar Anthropic tools are currently the number one choice among many computer coders.
*Hyperlocal Newsletter Firm — Powered by AI — Has a Million Subscribers: Newsletter maker Patch has come up with a new way to hyper-personalize newsletters with AI – even if only one person subscribes.
Observes writer Liz Skalka: “Give the site your zip code and it would produce a daily or twice-weekly newsletter customized for your town, for a minimum readership of one subscriber.
“The newsletters rely heavily on aggregation, automated event calendars and posts from Nextdoor (a hyperlocal social media network).”
*Grammarly Offering ‘Expert Reviews’ of Your Writing – Without Permission From Those Experts: AI proofreading tool Grammarly is now offering free reviews of your writing by prominent virtual academics, authors and similar experts – both living and dead – using AI.
Problem is, none of those experts granted Grammarly permission to use their thinking for such a purpose – which has more than a few people ticked.
Observes writer Miles Klee: “Instead of producing what looks like a generic critique from a nameless LLM, it lists a number of real academics and authors available to weigh in on your text. To be clear: Those people have nothing to do with this process.”
*The More You Chat, The More Errors You Get: New research finds that the more questions you ask in a specific chat, the more likely you’ll encounter errors and hallucinations.
Bottom line: Expect chatbots like ChatGPT-5.0 and newer to lose up to 33% accuracy if you engage in multi-message chats.
Even worse: Expect older AI engines to lose up to 39% accuracy.
*Another AI Tool Promises to Auto-Generate Press Releases: Add ‘Free AI Press Release Generator’ to the growing cadre of tools promising to create press releases for you with AI.
Unlike many AI writing tools, this one focuses entirely on press release writing and follows standard press release form and function developed by journalists.
*AI Agents May Soon Face Special Microsoft Fee: Microsoft is currently mulling adding a special fee for AI agents that use Microsoft products.
Observes writer Richard Speed: “The megacorp is considering a new Microsoft 365 subscription tier, informally dubbed E7, that would bundle Copilot and agent management tools.”
The reasoning: “As AI agents function as digital workers, they need identities, email accounts, Teams access, and policy controls – capabilities currently tied to user subscriptions,” Speed adds.
*Microsoft ‘Copilot Tasks:’ Redmond Takes Another Shot at AI Agents: Microsoft has rolled out a slightly different AI agent for consumer users of Copilot, designed to take a more active role in getting tasks done.
Observes writer Laurent Giret: “Microsoft described the feature as a To-Do list that does itself, done by an AI agent using its own computer and web browser in the background.”
Potential uses include creation of draft email replies each morning, general AI writing, online shopping, event planning, and more.
*Facebook Parent Meta Out With New AI Shopping Assistant: CEO Mark Zuckerberg’s empire has expanded a bit with a new, experimental AI shopping assistant.
Observes writer Mariella Moon: “At the moment, it’s reportedly only showing up on desktop browsers when select users visit Meta AI on the Web.”
Users prompting the assistant for shopping tips get back a virtual carousel of product images and pricing — as well as a link to where they can buy the product.
The AI is also able to customize suggestions based on gender and location if you share that personal data with Meta.
Share a Link: Please consider sharing a link to https://RobotWritersAI.com from your blog, social media post, publication or emails. More links leading to RobotWritersAI.com helps everyone interested in AI-generated writing.
–Joe Dysart is editor of RobotWritersAI.com and a tech journalist with 20+ years experience. His work has appeared in 150+ publications, including The New York Times and the Financial Times of London.
The post ChatGPT Now Clocking 900 Million Weekly Users appeared first on Robot Writers AI.
