🔓Anthropic's locked-down Mythos leaked
Anthropic's cybersecurity model Mythos, released April 10 to select partners under 'Project Glasswing' and deemed too powerful for public release, was leaked to a private Discord group within days of launch.
- —The model had triggered emergency meetings at the White House and other agencies due to its perceived capabilities
- —Bloomberg reports a Discord group tracking unreleased models accessed Mythos on release day and has used it regularly since
- —One member had vendor credentials through contract work, with leaked Mercor details helping the group locate and access Mythos online
- —The group told Bloomberg they do not use Mythos for cyberattacks or malicious activities, and claimed access to other unreleased models
Why it matters: The first alleged unauthorized use of an AI model dangerous enough to trigger government emergency meetings didn't come from China or Russia—it came from a random Discord group. As partner access grows and models get more dangerous, this problem will only compound.
🚀SpaceX stakes $60B on AI coding startup Cursor
SpaceX announced a major partnership with AI coding startup Cursor and locked in a $60B acquisition option, giving Elon Musk a shortcut into a race xAI has spent the year losing to Anthropic and OpenAI.
- —Cursor CEO Michael Truell said each Composer model release hit a compute ceiling, with SpaceX's Colossus now providing the needed power
- —Cursor is guaranteed $10B regardless, with the full $60B acquisition only happening if Musk exercises the option before year-end
- —xAI poached Cursor leads Andrew Milich and Jason Ginsberg last month, with Musk saying the startup "was not built right the first time around"
- —Cursor was set to raise $2B at a $50B valuation before the deal, with SpaceX holding off due to IPO process complications
Why it matters: Musk has tried and failed to build a frontier coding tool inside xAI, with Grok having no answer to Claude Code or Codex. This deal swaps in-house build for outside product to level up quickly, potentially turning Cursor from a compute-starved startup into a SpaceX-fueled coding lab.
✍️ Use this dictation strategy to write better docs
This two-step voice dictation system immediately makes you write better and improves itself over time.
- —Install Typeless for dictation, open a Codex/Claude Code session, and dictate: "Draft an outline for a short internal memo about [topic]"
- —Here's the key: tell the agent "Save this as the initial draft and do not edit it. Now create a separate working draft that we can revise"
- —Read it and use Typeless to add comments, pointing out inaccuracies, missing elements, and phrases that sound too generic or AI-like
- —Prompt: "Rewrite the draft using the comments I left. Write in my tone. Use my verbiage. No em dashes. Preserve core points, but cut anything that sounds generic."
Why it matters: Have an agent compare the untouched initial draft against final versions every week. It will extract your edits and update editorial rules for the future—this system improves current output while continuously learning your style.
💼ChatGPT's Codex-powered agents for teams
OpenAI introduced Workspace Agents, new Codex-powered shared bots designed to tackle multi-step team workflows autonomously across ChatGPT and Slack.
- —Workspace agents are pitched as an 'evolution' of 2023's solo-user GPTs, with old GPTs still working and a conversion tool coming soon
- —Backed by Codex in the cloud, agents can retain memory, call connected apps, and live in Slack or trigger on a schedule when users are offline
- —Inside OpenAI, sales reps use it for account research and follow-up drafts, while accounting runs it for journal entries and reconciliations
- —Custom agents can be created via ChatGPT and shared across teams, with restrictions on data usage, approvals, and permissions
Why it matters: OpenAI's enterprise push is no secret, and workspace agents solve a real problem—every team has scattered prompts and half-built workflows from the past two years, and few have unified them. The initial GPT Store didn't stick, but agentic upgrades and an enterprise shift could help this debut find a better fit.