Claude AI experienced a brief but widespread disruption on Monday evening when users reported persistent login errors and connection problems across the platform.
Anthropic acknowledged the incident on its system status page, saying the company was investigating elevated errors primarily affecting login and that it had identified the issue affecting Claude.ai and Claude Code logins.
According to DownDetector, reports spiked during the outage, with one live feed noting a surge of more than 4,000 reports before counts fell back to a few hundred, and another summary noting hundreds of users flagged issues.
DownDetector data cited in the reports showed a breakdown of affected services, with about 67 percent of users reporting problems with Claude Code, 14 percent with the Claude API, and 13 percent with the standard Claude Chat interface, as reported by the outage tracker.
Anthropic listed the services impacted by the incident as claude.ai, platform.claude.com formerly console.anthropic.com, the Claude API at api.anthropic.com, Claude Code, Claude Cowork, and Claude for Government.
The company said it was working to apply a fix and later reported that it had resolved the issue causing elevated login errors for Claude.ai and Claude Code, restoring access to its suite of tools.
Recurrence Context And Model Concerns
This outage follows other recent interruptions that affected Claude tools, reinforcing a pattern of short outages that often resolve quickly, according to live outage commentary.
One report noted that Anthropic recently released a preview of a new model called Claude Mythos, described as specialising in finding bugs and vulnerabilities, and that the company withheld a public release because of potential misuse risks, as reported by NDTV.
The same coverage said government officials and figures from international organisations have been holding meetings to evaluate cybersecurity risks tied to the model, and that Anthropic has faced scrutiny as it balances advanced capabilities with safety concerns.
Observers watching outage trends said the interruptions tend to be short, often lasting a few minutes to under half an hour, but high frequency of such blips has generated questions about whether backend capacity changes are needed to meet rising demand.
