# Andrej Karpathy Criticizes X, Musk Calls Algorithm Overhaul > Source: > Published: 2026-06-26 10:41:28+00:00 ### What happened India Today reports that **Andrej Karpathy** wrote on **X** that, after nearly two decades on the platform, "it has never been this toxic and Reddit-like," and that he has been spending more time on **Anthropic**'s Slack channel. Firstpost and India Today report the thread began while Karpathy was describing Anthropic's Claude Tag workflow, which he framed as moving AI toward a persistent, context-aware teammate integrated across tools, memory, and compute environments. Firstpost reports that **Elon Musk** responded by calling for a "complete overhaul" of X's recommendation algorithm. Digg notes Musk's comment did not include a timeline, scope, or measurable goals. ### Technical details Per India Today and Firstpost, Karpathy characterized Claude Tag as more than a simple chatbot or Slack integration; he described it as an assistant that can persist context across sessions and operate within organizational workflows. Launched June 23, 2026, Claude Tag is Anthropic's Slack-native async agent product: Claude joins as a team member with access to selected channels, tools, data, and codebases, and can be tagged in to delegate tasks asynchronously. Anthropic reports internally that Claude Tag writes 65% of its product team's code. It is in beta for Claude Enterprise and Team plans. ### Industry context Editorial analysis: Social platforms' ranking and recommendation systems have frequently been scrutinised for engagement-driven feedback loops that prioritise short-term reactions. Observers following similar debates note that platforms where ranking optimizes for engagement often see a rise in polarising or rapid-response content, which can degrade conversational quality and user experience. ### Context and significance Editorial analysis: The exchange matters because it pairs a senior AI practitioner's critique of platform dynamics with the platform owner publicly acknowledging algorithmic problems. For practitioners, this highlights two intersecting trends: the emergence of persistent, tool-connected AI assistants in enterprise workflows, and heightened scrutiny of how consumer-facing recommendation systems shape discourse. Neither the scraped reporting nor quoted posts provide technical blueprints for algorithmic change or product roadmaps. ### What to watch For practitioners: monitor whether X publishes technical notes, experimental results, or evaluation metrics tied to any algorithmic revisions; track official product documentation from **Anthropic** for Claude Tag that details persistence, context management, and privacy safeguards; and follow community signal changes such as engagement, moderation volume, or sentiment metrics reported by third-party observers. Also watch for any formal statements from platform teams that include timelines, ownership, or success metrics, since current reporting (Digg, Firstpost) indicates those elements are absent. ## Key Points - 1Karpathy publicly linked increased toxicity on **X** to the platform's recommendation dynamics, prompting owner-level acknowledgement. - 2Discussion of `Claude Tag` reframes enterprise AI as a persistent, workflow-integrated assistant rather than a transient chatbot. - 3Observers should watch for transparent algorithmic change metrics and formal product documentation from platforms and AI vendors. ## Scoring Rationale The story pairs a respected AI practitioner's public critique of platform recommendation dynamics with the platform owner's acknowledgement of the problem, and provides context on Anthropic's Claude Tag launch. Notable for practitioners interested in AI product direction and platform governance, but the exchange generates no new technical releases or measurable algorithmic changes. Practice with real Ride-Hailing data 90 SQL & Python problems · 15 industry datasets 250 free problems · No credit card [See all Ride-Hailing problems](/problems/datasets/mobility)