Anthropic has confirmed a major, system-wide artificial intelligence tools outage

Anthropic has confirmed a major, system-wide outage causing elevated error rates and widespread disruptions across its entire suite of artificial intelligence tools.

The incident, which began around 10:02 AM ET (14:02 UTC), is actively affecting Claude.ai, the developer API, the newly launched Claude Code platform, and the enterprise-focused Claude Cowork workflow environment.

Engineers at the high-profile AI firm have already identified the root cause of the platform-wide failure and are currently deploying an emergency fix to stabilize services.

At 10:53 AM Eastern Daylight Time (EDT), the company said, “A fix has been implemented and we are monitoring the results,” but users indicated that Claude was still inoperable twenty minutes later.

Anthropic is a leading American artificial intelligence safety and research company known for its family of large language models, including the Claude AI assistant and the agentic coding system, Claude Code.

Headquartered in San Francisco, Anthropic is currently valued at nearly $1 trillion and has confidentially filed for a U.S. Initial Public Offering on June 1.

Anthropic has been moving to secure computing capacity across multiple fronts as usage of its Claude models strains its infrastructure, and the company was in early talks to use Microsoft’s custom Maia 200 AI chips as part of a broader push to diversify its chip and compute partnerships that has also included arrangements with Amazon Web Services, Google, SpaceX, Broadcom, and CoreWeave.

According to published reports, Anthropic’s powerful Mythos AI model was able to break into “almost all” classified systems belonging to the National Security Agency (NSA) — one of the highest-ranking and most powerful intelligence agencies in the U.S. government — within hours during a controlled security evaluation.

Sen. Mark Warner, vice chair of the Senate Intelligence Committee, said Gen. Joshua Rudd, the head of the NSA and U.S. Cyber Command, briefed him on the model’s capability.

According to a June 14th report by The Economist, Rudd reportedly told Warner, the Fable 5 and Mythos 5 models “broke into almost all of our classified systems, not in weeks, but in hours.”

The quote initially went under the radar, then went viral about a week later across several social media platforms, generating false claims that Anthropic’s model “hacked the NSA.”

The breach actually occurred during an authorized internal test on June 11, in which Mythos was paired with other defensive tools under highly specific simulated environmental conditions.

The incident led to the June 12 U.S. government directive prohibiting all foreign nationals, including Anthropic’s own non-citizen employees, from accessing the Fable 5 and Mythos 5 models, due to national security concerns.

Anthropic responded by disabling both models globally, saying it could not enforce nationality-based access restrictions without denying everyone access to the systems.

Widespread Disruptions Hit Developers and Businesses

The sudden spike in server errors has ground daily operations to a halt for thousands of software developers, students, and enterprise clients who rely heavily on Anthropic’s ecosystem. Users attempting to interact with the platform are reporting a variety of critical failures, including:

• Failed Requests: Chat logs failing to load on the main Claude.ai web portal.
• API Timeouts: Severe latency and 500-level internal server errors breaking custom software integrations via the developer console.
• Broken Workflows: Complete request disruptions within specialized agentic tools like Claude Code and Claude Cowork.

Quick Identification, Fix Underway

According to the official Anthropic Claude Status Page, the infrastructure team moved quickly from an initial investigation to isolating the breakdown. The incident status has been updated to “Identified,” signaling that engineers know what went wrong behind the scenes and are actively rolling out a resolution patch.

While a concrete timeline for full recovery has not yet been announced, services are expected to stabilize gradually as the fix propagates across Anthropic’s global server networks.

Infrastructure Under Strain

This latest disruption underscores the ongoing scaling challenges faced by major artificial intelligence providers as massive agentic workflows and complex programming suites see rapid commercial adoption.

Observers note that the parallel processing demands of advanced coding assistants can place immense, sudden strain on backend server infrastructure.

For now, users are advised to monitor official channels for a green light and consider temporary local failovers or alternative models until normal operations resume.