Anthropic safety resignations hit 100M views on X as Threads leads in mobile DAU
The episode shows X still concentrates high-velocity AI insider discourse despite weaker mainstream engagement.
Two Anthropic employees kicked off a fast-moving AI safety news cycle on X, including a resignation thread that cleared 100 million views and a separate post warning of a “>10% chance” AI could kill all humans in the next decade. The burst is a clean example of X acting as the AI community’s real-time newswire even as Meta’s Threads leads X in daily active mobile users.
Anthropic’s Viral Safety Threads Show Where AI Narratives Break First
Anthropic researcher Jacob Coxon used X on a recent Tuesday to announce his resignation and publish a thread laying out safety concerns. The thread cleared 100 million views, turning what could have been an internal personnel story into a platform-scale distribution event.
In the same burst, another Anthropic employee posted that he thinks there is a “>10% chance” AI could kill all humans in the next decade. The pairing mattered because it compressed the usual pipeline. A resignation, a quantified extinction-risk claim, and the follow-on debate all propagated in the same place, at the same speed, before the wider information ecosystem had time to intermediate it.
Mechanically, this is what X is good at right now. It is the venue where model releases, research papers, demos, hiring moves, high-profile executive exits, and technical arguments ricochet through a single feed in real time. For traders who treat AI as a narrative driver across AI tokens and AI-adjacent infrastructure plays, the point is not whether the claims are right. The point is that the first draft of the story, including the most emotionally load-bearing lines, is increasingly written in public on X.
Threads Leads in DAU, but X Keeps the AI Insider Microphone
The mainstream engagement picture is not flattering for X. Similarweb data for January 2026 put Meta’s Threads at 141.5 million daily active mobile users versus X at 125 million. Daily active mobile users, or DAU, is the basic “how many people opened the app today” yardstick on mobile, and it is the metric advertisers and product teams tend to fight over.
That gap does not automatically decide where market-relevant AI discourse lands first. X’s current edge is concentration, not reach. The platform is described as a live “group chat” for the people closest to frontier AI labs, meaning the top-tier companies pushing the most advanced models and research where capability and safety debates originate.
The executive behavior in the same window points the same way. Nvidia CEO Jensen Huang launched his X account in July and posted in support of open AI models, meaning models released with weights or code, or broad access, so others can run and build on them. Meta CEO Mark Zuckerberg resumed posting on X in 2026 after leaving his account largely dormant since its creation in 2009. Apple’s new CEO John Ternus also joined to communicate major AI announcements.
X has also leaned into live video as an on-platform distribution layer for this audience, with shows like TBPN, Stabledash, and MTS Live. TBPN was acquired by OpenAI for a rumored “low hundreds of millions,” but the price is explicitly unconfirmed. The mechanism still matters: X is building places where AI insiders talk in public, and those clips and threads are easy to repackage into the day’s narrative.
Signals Traders Can Pull From X’s xAI Integration and Executive Posting
The xAI–X tie-up makes this less of a pure media story and more of a product loop. Musk’s xAI acquired X and said the companies would combine their data, models, computing power, and distribution. Grok, xAI’s model, can use a connected X account to read timelines and follow what’s happening on the platform, which means discourse spikes can become model inputs and product surface area.
Near-term, the clean signals are behavioral and disclosure-driven. Follow-up statements or additional resignations from Anthropic staff on X would extend the safety and scaling debate, and the packet does not specify the exact calendar date of the Tuesday posts. Any new disclosures from xAI or X about how Grok uses connected timelines, including expanded access or feature changes after the acquisition, would clarify how directly the feed is being productized.
The other datapoint to refresh is the platform split itself. Similarweb’s January 2026 DAU print is a snapshot, not a trendline. A widening Threads lead would reinforce X’s niche trajectory, while stabilization would suggest X is holding more mainstream attention than the narrative implies.
My Read: Treat X as the Early-Warning Feed for AI Risk and Policy Headlines
The part that decides this story is not whether Threads has more daily mobile users. It is whether the highest-velocity AI narratives, especially safety and policy shocks, still originate on X from people with direct proximity to frontier labs. The Coxon resignation thread clearing 100 million views, paired with an explicit “>10% chance” extinction-risk claim, is the kind of distribution event that can set the day’s framing before anyone else gets a turn.
If xAI keeps tightening the loop where Grok reads connected timelines and X keeps attracting executive posting for model and policy positioning, the setup starts to look structural rather than narrative-driven. In practical terms, X matters to markets if it remains the first place AI insiders publish the lines that everyone else has to price in afterward.
