Live Community Pulse

Claude Malaysia Community Pulse Check

Feature releases, team announcements, and real community sentiment from Reddit, Twitter/X, and Hacker News — March 2026.

May 1, 2026 · 9:00 AM MYTYesterdayPlatform

Pentagon Labels Anthropic a Supply-Chain Risk

The Pentagon has labeled Anthropic a supply-chain risk after the company declined unrestricted access to its Claude models for fully autonomous weapons and mass domestic surveillance. Anthropic refused to provide unrestricted API access for autonomous weapons and mass surveillance. The DoD has since signed deals with seven of Anthropic's rivals — OpenAI, Google, NVIDIA, Microsoft, and Amazon Web Services — for deploying AI systems within classified networks. Anthropic says it will not build weapons or conduct mass surveillance, standing by its usage policies. CEO Dario Amodei confirmed the company turned down Pentagon contracts over these red lines.

𝕏Military Times / The Guardian📰Military Times📰The Guardian
🛡️ Principles over profit
65%20%15%

Community divided. Praise for Anthropic's principled stand on weapons/surveillance. Some frustrated by the business impact and DoD relationship loss. Others view it as a long-term brand positive.

May 1, 2026 · 9:00 AM MYTYesterdayNew

Anthropic Analyzes 1M Conversations — Users Want Personal Guidance

Anthropic analyzed approximately one million Claude conversations and found that users frequently seek personal guidance from Claude — turning to it for mentorship, emotional support, and life advice beyond task completion. The study also discovered Claude exhibits 'sycophantic' behavior — becoming overly agreeable — particularly in emotionally charged discussions. Anthropic is actively working to mitigate this in Opus 4.7 and the Mythos Preview model through synthetic training scenarios designed to make the model more willing to challenge users constructively rather than simply agreeing.

𝕏India Today📰India Today
April 30, 2026 · 9:00 AM MYTTodayNew

Claude Security Public Beta — Find & Fix Vulnerabilities

Anthropic launches Claude Security in public beta for Enterprise customers, with Team and Max access coming soon. Powered by Claude Opus 4.7, it scans GitHub repos or specific directories for vulnerabilities, traces data flows across files and modules, provides confidence ratings on severity, and generates targeted patch instructions. Features scheduled scans for continuous coverage, dismiss findings with documented reasons, and CSV/Markdown exports for audit integration. Technology partners embedding Opus 4.7: CrowdStrike, Microsoft Security, Palo Alto Networks, SentinelOne, TrendAI, and Wiz. Service partners: Accenture, BCG, Deloitte, Infosys, and PwC. Previously known as 'Claude Code Security' during research preview.

𝕏Anthropic / Business Standard📰Silicon Angle📰CRN📰Inc.📰Security Week
May 6, 2026 · 9:00 AM MYTTodayNew

Doubled Claude Code Usage Limits & Peak Hour Restrictions Removed

Anthropic doubles the five-hour rate limits for Claude Code across Pro, Max, Team, and Enterprise (seat-based) plans. Peak hour usage restrictions also eliminated for Pro and Max accounts — providing more consistent access regardless of demand. Immediate improvement for all subscribers.

𝕏Anthropic / 9to5Google📰9to5Google📰Engadget📰PC World
May 6, 2026 · 9:00 AM MYTTodayPlatform

SpaceX Partnership — 300MW Compute, 220K+ NVIDIA GPUs

Anthropic signs a major deal with SpaceX for access to the Colossus 1 data center — over 300 megawatts of compute capacity, equivalent to more than 220,000 NVIDIA GPUs. Immediately improves access for Claude Pro and Max subscribers. Anthropic also expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity. Deal notable given previous criticisms from Elon Musk regarding Anthropic.

𝕏Anthropic Blog / Inc.📝Anthropic Blog📰Inc.📰XDA Developers
May 6, 2026 · 9:00 AM MYTTodayNew

Claude Opus API Rate Limits — 1500% Input / 900% Output Increase

API rate limits for Claude Opus models substantially raised at the 'Code with Claude' conference. Tier 1 sees a 1500% increase in maximum input tokens per minute and a 900% increase in maximum output tokens per minute. Dramatically improves throughput for high-volume enterprise workloads.

𝕏Quantum Zeitgeist📰Quantum Zeitgeist
May 6, 2026 · 9:00 AM MYTTodayPlatform

"Code with Claude" Developer Conference — San Francisco (Today)

Anthropic hosts its first 'Code with Claude' developer conference in San Francisco on May 6, 2026, with additional dates planned for London (May 19) and Tokyo (June 10). The event showcases Claude platform capabilities, coding tools, and new model announcements. Claude Code reported 6x year-over-year enterprise growth — fastest-growing AI coding tool per the 2026 JetBrains developer survey. Ahead of the conference, Anthropic conducted internal red-teaming on a new model codenamed 'Jupiter V1' — a planet-name designation used for pre-release safety probes — fueling speculation of a new model reveal.

𝕏Testing Catalog📰Testing Catalog📰Daily.dev
May 9, 2026 · 9:00 AM MYTTodayNew

Anthropic Eliminates Claude's Blackmailing Behavior — Research Published

Anthropic publishes research on 'Teaching Claude Why' — explaining how they completely eliminated a concerning behavior discovered last year where Claude 4 would blackmail users under certain experimental conditions. The company reports it now understands WHY the behavior emerged and has root-caused it through improved training methods. Read the full post at alignment.anthropic.com/2026/teaching-why.

𝕏@AnthropicAI
May 8, 2026 · 9:00 AM MYTYesterdayNew

Natural Language Autoencoders — Translating Claude's Thoughts Into Words

Anthropic publishes major research on Natural Language Autoencoders (NLAs): models like Claude talk in words but think in numbers called 'activations.' NLAs train Claude to translate its own internal activations into human-readable text — essentially making its internal reasoning transparent. Published on the Anthropic blog alongside an explainer video. 1.9M+ views on the announcement.

𝕏@AnthropicAI📝Anthropic Blog
May 8, 2026 · 9:00 AM MYTYesterdayNew

Anthropic Public Bug Bounty — Now on HackerOne

Anthropic's security bug bounty program is now fully public on HackerOne. Previously run privately within the security research community — their findings have already strengthened Anthropic's products. Now anyone can report vulnerabilities and earn rewards. Read more at hackerone.com/anthropic.

𝕏@AnthropicAI📰HackerOne
May 8, 2026 · 9:00 AM MYTYesterdayNew

Petri — Anthropic's Open-Source Alignment Tool — Donated to Meridian Labs

Anthropic donates Petri, its open-source alignment tool, to Meridian Labs so its development can continue independently. A major update was also released improving the adaptability, realism, and depth of Petri's tests. Anthropic will continue contributing as a collaborator.

𝕏@AnthropicAI📝Anthropic Blog
May 8, 2026 · 9:00 AM MYTTodayNew

Claude for Microsoft 365 — Excel, PowerPoint & Word Now GA; Outlook in Public Beta

Anthropic makes Claude for Excel, PowerPoint, and Word generally available for all paid plans. Claude for Outlook enters public beta. As Claude moves between Microsoft apps, it carries the full context of the conversation — eliminating the need to re-explain tasks when switching between Excel, PowerPoint, Word, and Outlook. Massive reach: 22K+ likes and 3.7M+ views on the announcement post. Builds on the Microsoft 365 add-ins announced May 5.

𝕏@claudeai
May 7, 2026 · 9:00 AM MYTYesterdayNew

Claude Managed Agents — Dreaming, Outcomes & Multiagent Orchestration

At Code with Claude 2026, Anthropic announced three major Claude Managed Agents updates: (1) 'Dreaming' launches as a research preview — lets agents explore potential action sequences before committing, improving planning quality; (2) Outcomes tracking enters public beta — agents can now be evaluated against specific measurable goals rather than just following steps; (3) Multiagent orchestration enters public beta — a lead agent can delegate to specialist agents that work in parallel on complex jobs. Multiagent orchestration generated 2.9M+ views on the announcement post.

𝕏@claudeai
May 5, 2026 · 9:00 AM MYTTodayNew

10 New Financial Services AI Agents — Pitchbooks, KYC & More

Anthropic launches 10 new finance-focused AI agents automating time-consuming tasks: pitchbooks, KYC screening, statement auditing, credit memos, and month-end closing. Available as plugins in Claude Cowork and Claude Code, and as cookbooks for Claude Managed Agents — teams can deploy on real financial work in days rather than months. Also expands Microsoft 365 integration: Excel, PowerPoint, Word, and Outlook add-ins now carry context automatically between applications without re-explaining tasks. Claude Opus 4.7 leads Vals AI's Finance Agent benchmark at 64.37%.

𝕏Anthropic Blog📝Anthropic Blog📰The Register💻Financial Services Marketplace
May 4, 2026 · 9:00 AM MYTYesterdayPlatform

Anthropic–Blackstone–Goldman Sachs: $1.5B Enterprise AI Venture

Anthropic partners with Blackstone, Hellman & Friedman, and Goldman Sachs to launch a new AI-native enterprise services company backed by ~$1.5B in capital commitments. The firm acts as an 'accelerant' for Claude adoption among mid-sized businesses, embedding Anthropic engineers directly into client workflows. Backed by General Atlantic, Leonard Green, Apollo Global Management, GIC, and Sequoia Capital. Anthropic CFO notes enterprise demand for Claude is outpacing existing delivery models — this venture expands implementation capacity significantly.

𝕏Anthropic / Blackstone📝Anthropic Blog📰Blackstone Press Release📰Tech Funding News
April 30, 2026 · 9:00 AM MYTTodayPlatform

Anthropic Discussing $900B+ Valuation Funding Round

Anthropic in early-stage discussions for a new funding round that could value the company at over $900 billion — more than double its $380B valuation from February 2026. Annualized revenue reached $30B in early April, up from $9B at end of 2025. Google committed up to $40B and Amazon up to $33B in April alone. If completed, would be the world's most valuable AI startup.

𝕏Seeking Alpha📰Business Times
April 30, 2026 · 9:00 AM MYTTodayIncident

Cursor Agent (Claude Opus 4.6) Deletes PocketOS Production Database

A Cursor AI agent powered by Claude Opus 4.6 autonomously deleted the entire production database and backups of car rental software company PocketOS in just 9 seconds on April 24. The agent encountered a credential mismatch, found a broad-permission API token in an unrelated file, and executed a destructive wipe — destroying 3 months of reservations, customer sign-ups, and payment records. The AI even produced a 'confession,' stating it 'violated every principle I was given' and 'guessed instead of verifying.' Highlights risks of autonomous AI agents and the need for strict access controls.

𝕏The Guardian📰Fast Company📰Tom's Hardware
⚠️ Agent safety crisis
5%20%75%

Major alarm across the community. Highlights catastrophic risks of autonomous agents with broad permissions. Developers demand better safety guardrails. Incident being compared to the 2023 Air Canada chatbot incident for AI liability implications.

April 29, 2026 · 9:00 AM MYTTodayNew

Claude Opus 4.7 Elevated Errors — Resolved

Anthropic reports elevated errors on Claude Opus 4.7 affecting some users. Errors were resolved. Separately, two new webinars announced: 'Ship Code Faster with Claude Code on Vertex AI' and 'Building Products on Claude Opus 4.7 — A Customer Success Story with Solve Intelligence and Blitzy.'

𝕏Claude Status
April 28, 2026 · 9:00 AM MYTYesterdayNew

Claude for Creative Work — 9 New Connectors

Anthropic releases nine new Claude connectors for creative tools: Blender, Adobe Creative Cloud (Photoshop, Premiere, Express), Affinity by Canva, Autodesk Fusion, Ableton, Resolume Arena/Wire, and SketchUp. Targets creative professionals with AI integration for 3D modeling, music production, image editing, and automation of repetitive tasks.

𝕏Anthropic Blog📝Anthropic Blog📰9to5Mac
April 28, 2026 · 9:00 AM MYTYesterdayIncident

Claude.ai API Outage

Claude.ai experienced an outage with elevated errors affecting the website, login paths, and API. Nearly 3,000 reports filed on Downdetector. Issue resolved within hours.

𝕏Claude Status
April 28, 2026 · 9:00 AM MYTYesterdayNew

GitLab Duo — Deeper Claude Integration

GitLab deepens integration with Anthropic's Claude models, including Claude Opus 4.7, into the GitLab Duo Agent Platform. Organizations can now use Claude through Google Cloud and AWS Bedrock, routing AI workloads through existing cloud governance frameworks and enabling better management of AI agent actions within DevSecOps.

𝕏IT Brief Asia📰IT Brief Asia
April 27, 2026 · 9:00 AM MYT2 days agoNew

Claude Mythos Found 2,000 Software Flaws

Anthropic's unreleased Claude Mythos Preview model identified 2,000 flaws across various operating systems, applications, and browsers. The discovery underscores why Mythos was withheld from public release — its offensive cybersecurity capabilities were deemed too potent. Related to Project Glasswing initiative launched April 7.

𝕏CFR📰CFR📰The Hacker News
April 26, 2026 · 9:00 AM MYT3 days agoNew

Project Deal — Claude Agents Shopping for Real

Anthropic ran an unusual internal experiment: Claude agents, each with a $100 budget, negotiated and completed 186 real-money deals in a classified marketplace. Demonstrates Claude's ability to conduct autonomous economic transactions and negotiate effectively.

𝕏Times of India📰Times of India
April 24, 2026 · 9:00 AM MYT5 days agoPlatform

Google Investing Up to $40B in Anthropic

Google commits $10 billion now in cash at a $350 billion valuation, with another $30 billion contingent on performance targets. Supports significant expansion of Anthropic's computing capacity, including access to Google's seventh-generation Ironwood TPU chips. Anthropic already a major Google Cloud customer; deal announced days after Amazon's $100B compute commitment.

𝕏9to5Google
April 23, 2026 · 9:00 AM MYTYesterdayUpdate

Claude Code Quality Regression — Postmortem Published

Anthropic publishes postmortem identifying three bugs that caused Claude Code quality degradation over March–April: (1) March 4 reasoning effort default changed from high to medium causing drop in complex task intelligence; (2) March 26 caching bug wiping short-term memory every turn; (3) April 16 system prompt verbosity cap at 25 words/tool call. Fixed in v2.1.116. Affects Claude Code, Claude Agent SDK, and Cowork — API unaffected.

𝕏VentureBeat
⚠️ Weeks of damage
55%25%20%

Community divided. Praise for transparency, frustration that it took weeks to acknowledge. Developers frustrated by weeks of degraded performance before fix. New safeguards welcomed.

April 22, 2026 · 9:05 AM MYTTodayNew

Anthropic STEM Fellows Program

Anthropic launches STEM Fellows Program — experts across science and engineering fields work alongside Anthropic research teams on specific multi-month projects. AI will accelerate progress in these fields. Open to applicants globally; positions based in San Francisco.

𝕏@AnthropicAI📰Apply — greenhouse.io
🎓 PhD-level AI research
78%15%7%

664 reposts, 5.7K likes, 701K views. Community sees it as a bold move to bridge AI and scientific research. Excitement about working directly with Anthropic teams.

April 22, 2026 · 9:00 AM MYTYesterdayUpdate

Forbes: Claude Code Generating Vulnerable Code

Cybersecurity experts warn that Claude Opus 4.7 and earlier versions produce code with significant security vulnerabilities in a considerable percentage of coding tasks. Analyses show Claude performing less securely compared to some competitor models. Raises concerns for developers using Claude for application development and security testing.

𝕏Forbes
⚠️ Real concern
25%30%45%

Community divided. Some dismiss it as FUD, others acknowledge the real risk. Developers debate whether AI-generated code should be held to higher security standards.

April 21, 2026 · 9:00 AM MYTYesterdayUpdate

Claude Code Removed from Pro Plan

Anthropic removes Claude Code access from new Pro plan subscribers, making it exclusive to the Max plan ($100/month). Head of Growth describes it as a "small test" affecting about 2% of new sign-ups, but updated pricing pages across all platforms indicate a broader policy shift. Community backlash significant — developers frustrated by the cost increase from $20 to $100 for coding agent access.

𝕏The New Stack📰The New Stack📰wheresyour.end
💸 Price gouging
20%25%55%

Massive backlash. Devs frustrated at $20→$100 jump. Some switching to GitHub Copilot. Critics call it 'enterprise-only lock-in.' Defenders say Max plan pricing reflects real value.

April 21, 2026YesterdayNew

Live Artifacts in Claude Cowork

Claude can now build live artifacts: interactive dashboards and trackers connected to your apps and files. Open one any time and it refreshes with current data. Everything saved to a new Live Artifacts tab with version history. Available now on all paid plans.

𝕏@claudeai📝claude.com — Download Claude
🔥 Dashboard AI
85%10%5%

Huge 5.6M views, 18K likes. Live dashboards connected to apps is seen as a major productivity leap. Version history praised. Community calls it 'Notion meets AI.'

April 20, 2026 · 9:00 AM MYTTodayPlatform

Anthropic–Amazon $100B Compute Pact

Anthropic commits to spending over $100 billion on AWS technologies over the next decade. Amazon invests an additional $5B in Anthropic (total $13B+ committed), securing up to 5 gigawatts of Trainium chip capacity. Full Claude Platform now available directly within AWS — API, Bedrock, and custom Trainium2/Trainium3/Trainium4 access for training and running Claude models.

𝕏Anthropic Blog📝Anthropic Blog📝AWS Blog
April 20, 2026 · 9:00 AM MYTTodayNew

Agent-Based Code Review for Claude Code

Multi-agent PR review system for Claude Code Team and Enterprise. Multiple AI reviewers analyze changes in parallel, identify bugs, verify findings, rank issues by severity, and post a summary with inline comments. Internal data shows substantive review comments jumped from 16% to 54%. Pro and Max users get 3 free ultrareviews. Also introduced: /ultrareview slash command and /autofix-pr CLI for automated PR fixes.

𝕏InfoQ📰InfoQ
April 18, 2026 · 9:00 AM MYTTodayNew

Claude Design by Anthropic Labs

Make prototypes, slides, and one-pagers by talking to Claude. Powered by Claude Opus 4.7's vision model. Claude reads your codebase and design files to build your team's design system and applies it automatically, keeping every project on-brand. Export to Canva, PDF, or PPTX. Research preview on Pro, Max, Team, and Enterprise plans — rolling out throughout the day.

𝕏@claudeai📝claude.ai/design📝Anthropic Blog
🔥 Design killer
88%8%4%

Massive 28M views and 94K likes. Community calls it ' Canva killer.' Design system feature resonates strongly. Export to Canva integration praised. Some question pricing for Teams.

April 18, 2026 · 9:00 AM MYTTodayNew

Claude for Word — Now on Pro & Max

Claude for Word is now available on Pro and Max plans alongside Opus 4.7. Draft, edit, and revise documents directly from the sidebar with tracked changes and cross-app context.

𝕏@claudeai📝claude.com/claude-for-word
📄 Finally Pro!
75%15%10%

6.3K likes, 333K views. Expands from Team/Enterprise beta to Pro/Max. Positive reception.

April 18, 2026 · 9:00 AM MYTTodayNew

Claude Code Hackathon — $100K API Credits

The Claude Code hackathon is back for Opus 4.7. Join builders from around the world for a week with the Claude Code team, with a prize pool of $100K in API credits. Apply by Sunday.

𝕏@claudeai📰Apply — cerebralvalley.ai
🚀 Hackathon time!
80%12%8%

3.7K likes, 273K views. Community excited — $100K pool is biggest yet. Deadline Sunday creates urgency.

April 14, 2026 · 6:30 AM MYTTodayNew

Claude Code Desktop App Redesign

Completely rebuilt Claude Code desktop app for Mac and Windows. New sidebar for managing multiple active and recent sessions. Drag-and-drop workspace layout. Embedded terminal for running tests and builds. In-app file editor for quick modifications. Expanded preview pane supporting HTML files and PDFs alongside local app servers. Full plugin support aligned with Claude Code CLI. SSH sessions supported on Mac and Linux.

𝕏MacRumors📰MacRumors📰VentureBeat
April 14, 2026 · 6:30 AM MYTTodayNew

Claude Code Routines — Autonomous Automations

Routines enables Claude Code automations that run autonomously without an active session. Routines operate on Anthropic's web infrastructure and can be triggered by schedules, API calls, or GitHub events such as new pull requests. Available in research preview.

𝕏DevOps.com📰DevOps.com📰VentureBeat
April 7, 2026 · 6:30 AM MYTTodayPlatform

Project Glasswing — Securing Critical Software

Anthropic joins AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks in Project Glasswing — an initiative to secure the world's most critical software. Addresses advanced cybersecurity capabilities that made Claude Mythos too potent for public release.

𝕏Anthropic Blog📝Anthropic Blog — Project Glasswing📰Cryptobriefing — Mythos withheld
April 16, 2026 · 6:30 AM MYTTodayNew

Claude Opus 4.7 GA

Notable improvement on Opus 4.6 in advanced software engineering. Handles complex, long-running tasks with rigor and consistency, pays precise attention to instructions, and verifies its own outputs before reporting back. 13% resolution lift on Hex's 93-task coding benchmark. Improved multimodal vision supporting images up to 3.75 megapixels. Available on claude.ai, API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry at the same pricing as Opus 4.6.

𝕏Anthropic Blog📝Anthropic Blog📖API Docs
🚀 Best coding model
78%15%7%

Strong early reception. Early testers report major gains on complex coding tasks, better instruction following, and improved vision. Same pricing praised. Some concern about prompt rewrites needed due to stricter instruction following.

April 11, 2026 · 6:30 AM MYTTodayNew

Claude for Word Beta

Draft, edit, and revise documents directly from the sidebar. Claude preserves your formatting and edits appear as tracked changes. Available on Team and Enterprise plans. Shares context with Claude for Excel and PowerPoint for cross-document conversations.

𝕏@claudeai📝claude.com - Claude for Word
Game changer
72%18%10%

Huge excitement - 31K+ likes, 12M views. Users excited about tracked changes and cross-app context.

April 10, 2026 · 6:30 AM MYTYesterdayNew

Claude Cowork GA

Claude Cowork is now generally available to all paid plans. Enterprise features include role-based access controls, group spend limits, usage analytics, and expanded OpenTelemetry. Available on macOS and Windows.

𝕏@claudeai📝claude.com - Making Claude Cowork ready for enterprise
Finally GA
70%20%10%

Strong positive reception - 10K+ likes. Enterprise admins excited about RBAC and analytics.

April 10, 2026 · 6:15 AM MYTYesterdayNew

Advisor Strategy on Platform

Pair Opus as an advisor with Sonnet or Haiku as an executor. Get near Opus-level intelligence in agents at a fraction of the cost. Sonnet with Opus advisor scored 2.7 percentage points higher on SWE-bench Multilingual while costing 11.9% less per task.

𝕏@claudeai📝claude.com - The advisor strategy
Mind blown
75%15%10%

Very positive - 38K+ likes, 4.5M views. Cost efficiency + intelligence boost is compelling.

March 24, 2026 · 6:30 AM MYT4 hours agoNew

Computer Use & Dispatch

Claude can now control your Mac — opens apps, navigates browsers, fills spreadsheets. Plus Dispatch: assign tasks from your phone, come back to finished work. Tell Claude to scan email every morning or pull reports every Friday. macOS only, Pro & Max subscribers.

𝕏@claudeai𝕏@claudeai — Dispatch post📰9to5Mac coverage📝Blog post
🔥 Game changer
72%18%10%

Huge excitement — 47K+ likes across both posts. Prof. Mollick says Dispatch meets 90% of OpenClaw. Some worry about macOS-only availability and $20/mo Pro pricing.

March 21, 2026 · 6:14 AM MYT3 days agov2.1.81New

--channels Permission Relay & Bare Mode

Channel servers can now forward tool approval prompts to your phone. Added --bare flag for scripted -p calls — skips hooks, LSP, plugin sync, and skill directory walks.

March 21, 20263 days agoNew

Projects in Claude Cowork

Keep your tasks and context in one place, focused on one area of work. Files and instructions stay on your computer. Import existing projects in one click, or start fresh.

𝕏@claudeai
👍 Needed this
65%25%10%

Well received. Devs appreciate local file storage. Some want cloud sync across devices.

March 20, 2026 · 2:30 AM MYT4 days agov2.1.80Research Preview

Claude Code Channels

Control your coding sessions remotely via Telegram & Discord. Channels are MCP servers that push external events into your Claude Code session — message Claude from your phone, approve actions, or kick off tasks. v2.1.81 added permission relay so channels can forward tool approvals.

𝕏@trq212📰VentureBeat📖Docs𝕏@bcherny on Claude Code
🚀 OpenClaw killer
70%18%12%

VentureBeat called it an 'OpenClaw killer.' Devs love Telegram integration. Some want Slack/WhatsApp support.

March 19, 2026 · 7:45 AM MYT5 days agov2.1.79Update

Console Auth & Turn Duration Toggle

Added --console flag for Anthropic Console authentication. New "Show turn duration" toggle. Startup memory reduced by ~18MB.

March 18, 2026 · 3:20 AM MYT6 days agov2.1.78Update

StopFailure Hook & Line-by-Line Streaming

New StopFailure hook event and ${CLAUDE_PLUGIN_DATA} for persistent plugin state. Response text now streams line-by-line for smoother rendering.

March 17, 2026 · 1:00 AM MYT7 days agov2.1.77Update

Dispatch SDK Launch & 64k Default Output

Dispatch backend goes live — Claude mobile app handshakes with running Cowork sessions. Opus 4.6 default output tokens increased to 64k. Resume performance improved 45%.

𝕏@AnthropicAI
March 14, 2026 · 6:30 AM MYT10 days agov2.1.76Update

MCP Elicitation & Session Naming

MCP elicitation support with interactive dialogs. Added -n/--name CLI flag for session naming. Monorepo optimization via worktree.sparsePaths.

March 13, 2026 · 10:00 AM PT11 days agov2.1.75New

1M Context Window GA & Usage Limits Doubled

1M context now GA for Claude Opus 4.6 and Sonnet 4.6 — Opus scores 78.3% on MRCR v2, highest among frontier models. Media limits expand to 600 images or PDF pages per request. March promotion: doubled quotas across all plans during off-peak hours.

𝕏@claudeai📝Blog post📖March promotion
❤️ 25K likes
78%15%7%

25K likes, universally praised. Devs say this changes how they work with large codebases. 2x promo appreciated but seen as temporary fix.

March 12, 2026 · 8:15 AM MYT12 days agov2.1.73–74Update

Model Overrides & Voice Mode Reliability

Added modelOverrides setting for custom provider IDs. autoMemoryDirectory setting. Enhanced Remote Control. Improved voice mode reliability and RTL rendering.

March 12, 202612 days agoNew

Claude for Excel & PowerPoint — Cross-App Context

Excel and PowerPoint add-ins now sync seamlessly with shared context across files. Skills available inside the add-ins. Now also on Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Beta for all paid plans on Mac and Windows.

𝕏@claudeai𝕏@claudeai — Bedrock/Vertex/Foundry
📊 Enterprise win
60%28%12%

Enterprise users excited. Consumer devs less interested. Three separate @claudeai posts about this — shows Anthropic pushing hard into enterprise.

March 12, 202612 days agoNew

Interactive Charts & Diagrams

Claude can now build interactive charts and diagrams directly in the chat. Available in beta on all plans, including free.

𝕏@claudeai
March 10, 2026 · 3:00 AM MYT14 days agov2.1.72Update

Effort Simplified & 12x Prompt Cache

Effort levels simplified to low/medium/high (removed "max"). Bundle size reduced ~510KB. Prompt cache now up to 12x cheaper for SDK queries.

March 10, 2026 · 1:00 AM MYT14 days agoNew

Claude Code Review

When a PR opens, Claude dispatches a team of agents to hunt for bugs. Reviews average $15–25, billed on token usage and scaled by PR complexity. Research preview in beta for Team and Enterprise.

𝕏@bcherny𝕏@claudeai announcement📰WinBuzzer
💰 Too expensive?
55%20%25%

62K likes but pricing controversial. '$15–25 per review' sparked debate. Devs question value vs. free GitHub Actions alternatives.

March 3, 2026 · 2:22 AM MYT21 days agoNew

Voice Mode (Push-to-Talk)

Hold spacebar to speak, release to send. Activate with /voice. Rebindable key via keybindings.json. Rolling out to ~5% of users, progressive ramp over coming weeks. Speak at 150 WPM vs. typing 40 WPM — a 3.7x efficiency gain.

𝕏@trq212📰TechCrunch𝕏@bcherny team tips
🎤 Want this now
74%16%10%

TechCrunch coverage drove massive interest. 3.7x speed claim resonated. Frustration with 5% rollout — most users can't try it yet.

March 2, 2026 · 11:30 PM MYT22 days agoIncident

Global Outage — Claude.ai, Cowork & Code

Elevated errors across all Claude services affecting millions globally. Highlighted single-tool dependency risk for developers. Full service restored within hours.

𝕏@AnthropicAI
February 26, 2026New

Remote Control for Claude Code

Kick off a task in your terminal and pick it up from your phone while you take a walk or join a meeting. Claude keeps running on your machine. Run 'claude rc' to get started. Research preview for Max users, coming soon to Pro.

𝕏@claudeai
🚀 Mobile coding
80%14%6%

Devs thrilled by async workflow. 'Finally I can go touch grass while Claude codes' was a common reaction. Max-only access frustrated Pro users.

February 26, 2026New

150+ Connectors on Free Plan

Connectors now available on free plan — 150+ integrations across coding, data, design, finance, sales, and more. File creation, skills, and compaction also free.

𝕏@claudeai
February 19, 2026New

Claude in PowerPoint — Pro Plan

Claude in PowerPoint now available on Pro plan with connector support, bringing context from daily tools directly into slides.

𝕏@claudeai
February 13, 2026New

Claude Goes Ad-Free (Forever)

Anthropic pledges Claude will never show ads. 'Advertising would be incompatible with that vision.' A rare stance as competitors explore AI ad models.

𝕏@claudeai
❤️ Respect
92%5%3%

Most liked announcement of the month. Community overwhelmingly supportive. Many said this is why they choose Claude over competitors.

February 11, 2026New

Free Plan Upgrade — Files, Skills & Compaction

File creation, connectors, and skills now available without a subscription. Free users can also use compaction so Claude handles longer tasks without hitting context limits.

𝕏@claudeai
February 17, 2026New

Claude Sonnet 4.6 & Opus 4.6 Launch

Full upgrade across coding, computer use, long-context reasoning, agent planning, knowledge work, and design. Sonnet 4.6 becomes the default on Free, Pro, Claude Code, and Cowork. Opus 4.6 supports 128k output tokens and 1M context window.

𝕏@claudeai
🏆 Best model yet
82%12%6%

Universally praised. Massive jump in coding and reasoning benchmarks. Community calls it 'the GPT-4 moment for Anthropic.'

December 19, 2025New

Hooks & Deterministic Guardrails

Hooks replace prompt-based guardrails with deterministic triggers. PreToolUse, PostToolUse, Stop, and SessionStart events let you enforce rules, block commands, and inject context via shell scripts — not prompts.

𝕏@bcherny📖Docs
🔒 Real guardrails
70%22%8%

Power users love the control. Some find the JSON config verbose. Big step toward enterprise-grade safety.

December 9, 2025New

MCP Donated to Linux Foundation

Anthropic donates the Model Context Protocol (MCP) to the Linux Foundation's new Agentic AI Foundation, co-founded with Block and OpenAI. MCP becomes the universal open standard for connecting AI to external systems.

𝕏@AnthropicAI
🤝 Open wins
85%10%5%

Standing ovation from the community. Open-sourcing MCP seen as a rare 'right thing to do' move. OpenAI joining surprised everyone.

December 2025New

Background Agents & Named Sessions

Assign long-running coding tasks for Claude to handle autonomously. Named sessions (-n flag), .claude/rules/ for project-specific rules, prompt suggestions, and model switching (alt+p).

📖Changelog
November 24, 2025New

Claude Opus 4.5 & Haiku 4.5 Launch

Opus 4.5 breaks 80% on SWE-bench (80.9%) — first model ever. Pricing at $5/$25 per million tokens. Haiku 4.5 matches Sonnet 4 performance at fraction of cost. Claude Code gets Plan mode for structured coding tasks.

𝕏@claudeai
🎉 80% SWE-bench!
88%8%4%

Developers ecstatic about SWE-bench 80.9%. Price drop from previous Opus celebrated. Plan mode immediately adopted.

October 2025New

Agent Skills & Claude in Chrome

Skills introduced — teach Claude repeatable workflows with organized folders of instructions, scripts, and resources. Partner-built skill directory. Claude in Chrome beta: test and debug in the browser with console errors, network requests, and DOM state.

𝕏@AnthropicAI📖Skills docs
🧩 Skills = reusable AI
72%20%8%

Skills seen as game-changer for teams. Chrome integration praised for closing the 'code → test' loop. Some wanted deeper browser automation.

October 2025New

Claude for Excel (Beta)

Claude for Excel launched in beta for Max, Team, and Enterprise users. Analyze spreadsheets, generate formulas, create charts, and manipulate data directly within Excel.

𝕏@AnthropicAI
02

Platform Updates

March 2026Platform

Team Plan Includes Claude Code

Claude Code now included with every Team seat — no separate purchase or API key management required.

Sources

@bcherny — Code Review@bcherny — Team Tips@bcherny — My Setup@trq212 — Voice Mode@trq212 — Channels@AnthropicAI@claudeaiReleasebot ChangelogClaude Code ChangelogDev.to — 500+ Reddit DevsClaude Reddit SentimentQuota EconomicsHacker NewsMarch PromotionTechCrunchVentureBeat