A quiet week on the feature front, but a loud one on the policy front. On Friday, June 26, the US government cleared Anthropic to release Mythos 5 — one of its most capable models — to a limited set of companies and federal agencies, easing an export block it had ordered earlier in June. Here is what happened and what it means for the models behind Cowork.
A Limited Green Light for Mythos 5
Earlier this month, the US government directed Anthropic to suspend access to its newest frontier models after the most powerful of them raised cybersecurity concerns. On June 26, that stance softened: the administration granted Anthropic permission to release Mythos 5 to a limited group of companies and government agencies, revising the earlier license requirements rather than lifting the restriction entirely.
Mythos 5 is not a model most Cowork users will touch directly. The point is the direction of travel: Anthropic's most capable systems are moving into regulated and government settings under tighter, case-by-case access rules. For everyone building on Claude, it is a useful signal of how the frontier is being governed even as the underlying capability keeps advancing.
A Brief Wobble on June 23
The week also brought a reminder that frontier infrastructure is still infrastructure. On June 23, Claude experienced a widespread outage that left many users unable to connect; Anthropic later reported the incident as resolved. Service disruptions like this are infrequent, but they are worth planning around if you lean on Claude for time-sensitive work.
Why It Matters for Cowork Teams
Cowork is only as available and as capable as the Claude models underneath it. The Mythos 5 decision shows regulators and Anthropic working out how the most capable models reach high-stakes users — a backdrop that will shape what lands in everyday products over time. The June 23 outage is the more practical reminder: when Claude runs long, unattended work for you, it helps to design tasks so a hiccup is recoverable rather than catastrophic.
In short, neither item changes a setting in your Cowork workspace today, but both are worth knowing as you decide how much to trust Claude with — and when.
Tip of the Week: Let Cowork Run It on a Schedule
If you find yourself asking Claude for the same thing every Monday — a status roundup, a metrics pull, a research sweep — stop doing it by hand. Cowork can run recurring work on a schedule and hand you the finished output when it is done.
❌ Weak
Open Cowork every Monday and re-type the same prompt to pull last week's numbers into a summary.✅ Strong
Ask Cowork to set up a scheduled task that runs the report every Monday morning and saves the result for you — write the instructions once, review the output each week.The rule of thumb: if a task is predictable and repeats on a calendar, describe it once and let a schedule do the running.
Worth Reading This Week
- → US allows Anthropic to release Mythos 5 to some companies and agencies — CNBC — How the government revised its export restriction to permit a limited release.
- → US government allows Anthropic limited release of Mythos — CNN — Context on the cybersecurity concerns behind the earlier export block.
- → Claude was down for many — Anthropic says the outage is resolved — TechRadar — A timeline of the June 23 disruption and recovery.
- → Anthropic Newsroom — The company's own announcements, including statements on the model availability decisions.
No new buttons in Cowork this week, but plenty shifting around it. We will keep watching the model availability story — and the features it eventually unlocks for the rest of us.
Until next time.
— The ClaudeCowork.com team