OpenAI has reported an ongoing service incident involving elevated errors across ChatGPT and Codex, with the company confirming that it is investigating degraded performance affecting the two platforms.
The issue was marked as “Investigating,” indicating that engineering teams had identified abnormal behavior but had not yet published a confirmed root cause, mitigation timeline, or full technical explanation.
According to the OpenAI status page, the incident affects ChatGPT and Codex, two widely used services for AI-assisted conversations, software development, code generation, and programming workflows.
The status notice states that OpenAI is investigating “the issue for the listed services,” while the affected-components section shows 15 ChatGPT components and four Codex components listed under the ongoing event.
The reported disruption may result in users encountering failed requests, unexpected error messages, delayed responses, interrupted conversations, or difficulty accessing particular product functions.
Developers and security teams that depend on Codex-powered coding workflows could also experience reduced reliability during code generation, debugging, automation, or repository-analysis tasks. However, OpenAI has not confirmed which specific features, regions, account types, or underlying systems are experiencing the highest error rates.
The incident was publicly listed at 08:28 PM on September 3, 2026, and remained under investigation several minutes later. At the time of the update, OpenAI’s system-status dashboard continued to display overall service-availability figures for its broader product categories, including APIs, ChatGPT, and Codex.
Those historical uptime figures should not be interpreted as confirmation that all individual features are currently operating normally, as a partial outage can affect selected components while the wider service remains available.
The incident also follows recent industry attention on service disruptions affecting Anthropic’s Claude AI platform. Cyber Security News previously reported on a Claude AI outage that affected access to Anthropic’s services, highlighting how outages at major AI providers can disrupt developers and enterprise users who depend on cloud-hosted models.
For enterprise users, the event highlights the operational risk of relying on a single external AI platform for time-sensitive work. Security operations centers, development teams, and content teams increasingly use generative AI tools for summarizing alerts, drafting incident reports, reviewing code, analyzing logs, and accelerating vulnerability research.
A temporary service degradation can slow those workflows, particularly where AI-based processes are integrated into ticketing, CI/CD, developer tooling, or internal knowledge systems.
Organizations using ChatGPT or Codex for critical business functions should monitor the official status page for updates rather than relying solely on social-media reports or isolated user complaints. Teams may also want to activate fallback processes, such as local development tools, manual review procedures, alternate automation paths, and cached documentation, until OpenAI confirms that service performance has returned to normal.
At present, OpenAI has not attributed the elevated errors to a cyberattack, data breach, infrastructure failure, or third-party dependency issue. The available advisory only confirms degraded performance and an active investigation.
Users are advised to retry failed requests cautiously, avoid repeatedly submitting sensitive content, and preserve error timestamps or request identifiers if they need to report persistent issues to support teams.
Learn 7 Metric-Gated AI SOC Deployment Phases – Download Free AI SOC Deployment Playbook 2026.
