Claude Ultra replaces Max 20x and drops verified accounts’ five-hour limit
Anthropic renames Max 20x to Ultra at $200 a month. Verified accounts lose the five-hour ceiling, but weekly limits remain unchanged.
Anthropic
Anthropic news covering Claude, coding tools, AI research, safety, company strategy, and policy, with clear context and links to the evidence.
Claude Code 2.1.314 adds write-tool boundaries and enterprise routing
What the new controls govern, where the documentation stops, and why file editing and shell execution need separate scrutiny.
Claude’s new cost analytics API trades speed for richer reporting
Anthropic’s October 29 beta adds unified reporting, but developers should check data freshness and migration behavior before switching dashboards.
Claude mobile update brings CarPlay and better tablet layouts
New mobile controls reduce app-switching, while voice limits and sharing restrictions still matter.
Anthropic expands AI for Science beyond individual grants
Institutional partnerships sit alongside existing grant applications and a separate, limited Team promotion.
Claude Enterprise adds smart reports, surveys and shared budgets
Three Enterprise features reach general availability, with different eligibility, privacy and spending rules.
Claude adds legal artifacts, with source links and sharing limits
Claude’s new legal gallery offers reusable workflows. Source checks, connector access and sharing rules still need attention.
Artificial Analysis Index v5 changes the test, not just the ranking
A rebuilt benchmark mix and a separate Progress Index change how to read AI scores. Reasoning effort and task cost matter as much as the headline ranking.
Anthropic’s GSA deal expands discounted Claude access beyond federal agencies
The new OneGov agreement offers a 40% usage discount with no minimum order. Direct purchases and future cloud-provider discounts need to be distinguished.
Anthropic files for an IPO with a split control structure
Anthropic has filed its S-1 for a proposed Nasdaq offering. The filing separates public shares, founder voting power and the Long-Term Benefit Trust's board rights.
Agent Skills moves to AAIF with a roadmap for shared governance
Anthropic is transferring Agent Skills to AAIF. The format remains open, while client permissions, dependencies and workflow testing still matter.
Claude adds organization controls for web fetch
Anthropic’s October 20 update adds organization web fetch controls. API requests and agent sessions handle disabled access differently. Managed Agents also checks redirect destinations.
Claude Security adds continuous scans and tracks fixes
Claude Security adds continuous scans, fix tracking and integration controls in its Enterprise public beta. Migration and usage credits have limits.
Claude Duet preview access, privacy and approval controls
Claude Duet’s US preview shares Claude usage limits. Learn what disconnecting changes, how approvals work and what to check before delegating a task.
Claude API update brings breaking SDK migration changes
Anthropic releases a new API version and major client updates. Request validation, gateway handling and older Bedrock models need attention before migration.
Proteins designed with Claude bind scorpion toxins
Anthropic reports 17 qualifying binders among 211 designs. Toxin neutralization and animal efficacy remain untested.
Claude adds customer level API enforcement and beta webhooks
Anthropic adds end customer attribution and enforcement notifications, with separate platform support and access requirements.
Anthropic study tracks a shrinking share of junior technical hires
New research examines hiring, agent supervision and differing expectations about the future of work.
Anthropic reconstructs the origins of NIST curve seeds
Anthropic traces historical curve seeds to timestamps. The findings explain parameter origins while leaving several questions open.
Claude Platform adds tracing for API requests and agents
Anthropic’s October 14 tracing beta connects API and SDK activity. Developers need to choose content settings, verify delivery and avoid counting tokens twice.
Anthropic expands METR access for risk assessments
Six evaluators gain internal access with publication rights and exclusions.
Simons launches $15 million effort to help mathematicians assess AI
The partnership will support mathematical review and provide AI resources. The announcement leaves application details and access terms unspecified.
Claude plans voting information in chat for US midterms
Anthropic plans Democracy Works answers in Claude, with rollout and results details still to come.
Claude Code 2.1.299 changes file mentions and idle context handling
What to check before using the new file mention control and idle compaction behavior in a long coding session.