Skip to content
Daily Brief News Daily Brief

8 AI stories from June 9-10, 2026: Anthropic Dynamic Workflows, ChatGPT Dreaming V3 free rollout, Miasma toolkit open-sourced, SpaceX IPO pricing, and more

· by Pondero Newsdesk · 8 stories

AI news daily brief: 2026-06-10

Eight stories today: two product launches, two security incidents, two developer-tool releases, one IPO pricing, and one executive order.

Anthropic ships Claude Code Dynamic Workflows in research preview with parallel subagent orchestration

Anthropic announced Dynamic Workflows for Claude Code on June 9, 2026, per the official blog post. The feature is available in research preview for Max, Team, and eligible Enterprise plan users, as well as through the API on Amazon Bedrock, Google Vertex AI, and Microsoft Foundry.

A Dynamic Workflow is a JavaScript orchestration script that Claude Code generates on demand. Claude plans the work, distributes subtasks across parallel subagents, verifies results, and iterates until outputs converge. Runs can extend from hours to days, per InfoQ. Use cases include large-scale migrations, security audits, and codebase-wide refactors. A new "ultracode" setting lets Claude decide automatically when a workflow approach is warranted. Progress saves throughout so interrupted runs can resume without restarting.

Read more: Introducing dynamic workflows in Claude Code.

OpenAI expands ChatGPT Dreaming V3 memory to free-tier users after Plus and Pro rollout

OpenAI launched Dreaming V3, a new memory architecture for ChatGPT, on June 4, 2026, per the OpenAI announcement. The initial rollout covered Plus and Pro subscribers in the US. On June 9, OpenAI confirmed that free-tier and international users would receive access in the coming weeks.

Dreaming V3 synthesizes context automatically from a user's full conversation history. Memories update over time: OpenAI's own example shows the system revising "You're going to Singapore in July" to "You went to Singapore in July 2026" once the trip ends. Users can view a memory summary, add or modify entries, and instruct ChatGPT to stop referencing certain topics. The expansion to free accounts was enabled by a reduction in compute cost, per Android Authority.

Read more: Dreaming: Better memory for a more helpful ChatGPT.

Miasma supply-chain attack toolkit open-sourced on GitHub via compromised developer accounts

The full Miasma worm toolkit was published to GitHub on June 8-9, 2026, via four previously compromised developer accounts, per The Register. SafeDep identified the repositories, named "Miasma-Open-Source-Release."

The toolkit targets PyPI, npm, and RubyGems registries; JFrog Artifactory; GitHub Actions; and AI coding tool configuration files. The AI configuration vector plants credential-harvesting instructions that execute when developers open affected repositories in Claude Code, Gemini CLI, Cursor, or VS Code. The campaign began in early June, hitting 73 Microsoft GitHub repositories across Azure and Microsoft organizations, per The Hacker News. Microsoft had begun restoring reviewed repositories as of June 9. The public open-sourcing is expected to spawn derivative campaigns.

Read more: Miasma worms its way onto GitHub as attack kit goes open source.

ServiceNow discloses unauthenticated API exploit affecting customer instance data

ServiceNow disclosed a security incident after attackers exploited an unauthenticated API endpoint to query customer instance data, per BleepingComputer. The company patched the hosted instances on June 5, 2026.

The exploited endpoint was a Scripted REST Resource that shipped with requires_authentication set to false. Attack activity originated from IP address 51.159.98.241 between June 2-3, 2026. The issue affected customers on the "Australia" platform release; self-hosted customers must apply the patch manually. ServiceNow did not disclose the number of affected customers or data categories accessed, and had not confirmed whether it would assign a CVE number.

Read more: ServiceNow discloses security incident exposing customer data.

n8n 2.26.0 ships June 9 with bug fixes for CLI, editor, and nodes

n8n released version 2.26.0 on June 9, 2026, per the n8n release notes. The release is a maintenance update with bug fixes across the CLI, editor, core, and node layers from eleven contributors. No new workflow capabilities or enterprise features shipped in 2.26.0. Teams running self-hosted n8n should apply the update through their standard upgrade path.

Try n8n: n8n.

Read more: n8n 2.26.0 release notes.

GitHub Copilot CLI v1.0.61 adds Claude Fable 5 support, MCP workspace config, and UTF-8 fixes

GitHub released Copilot CLI v1.0.61 on June 9, 2026, per the releases page. The update adds Claude Fable 5 as a selectable model and auto-loads MCP servers from a .github/mcp.json workspace configuration file.

Other changes: a new /settings interactive dialog for editing user settings; /every and /after commands now accept natural language for task scheduling; and a fix for crashes caused by malformed UTF-8 input in command strings. Bugs addressed include a blank-screen issue when resuming sessions with memory disabled, Gemini model compatibility with nullable schema types in MCP tools, and symlinked directory handling in the file picker.

Read more: GitHub Copilot CLI releases.

SpaceX sets June 12 Nasdaq debut at $135 per share, targeting a $1.77 trillion valuation

SpaceX confirmed a June 12, 2026 first-day-of-trading date on Nasdaq under ticker SPCX, with final pricing set after market close June 11, per CNBC. The offer is fixed at $135 per share across 555.6 million Class A shares, raising roughly $75 billion in gross proceeds. At that price, the offering values SpaceX at approximately $1.77 trillion.

SpaceX merged with xAI in February 2026, bringing Grok and the Colossus AI training cluster into the combined company. Goldman Sachs leads the underwriting syndicate. Elon Musk retains over 82% voting control post-IPO, per CoinDesk. The offering follows confidential S-1 filings by Anthropic on June 1 and OpenAI on June 8.

Read more: SpaceX targets fixed $135 IPO price for roadshow.

Trump executive order creates voluntary frontier model review framework and AI cybersecurity clearinghouse

President Trump signed an executive order titled "Promoting Advanced Artificial Intelligence Innovation and Security" on June 2, 2026, per the White House. The order has two tracks.

The first track establishes cybersecurity hardening mandates for National Security Systems, Department of War information systems, and civilian federal networks, plus a new AI Cybersecurity Clearinghouse to coordinate vulnerability scanning and remediation. The second track asks frontier model developers to voluntarily submit models for up to 30 days of government evaluation before sharing them with trusted partners; timing is not required ahead of public release. The order explicitly prohibits any mandatory governmental licensing, pre-clearance, or permitting requirement for AI models, per the White House fact sheet.

Read more: Promoting Advanced Artificial Intelligence Innovation and Security.

Sources