← Latest · Archive

SEV-3OpenAI
2 sources standard

OpenAI announced on 20 July 2022 that DALL·E, its text-to-image generation system, had entered public beta after operating in private preview since January 2021 [source]. The company stated that over one million users would be granted access from the waitlist, with plans to invite approximately 1,000 additional users per week.

The announcement marked a significant expansion in availability but introduced new operational constraints. OpenAI implemented a credit-based pricing system where users receive 50 free credits in their first month and 15 free credits monthly thereafter, with additional credits available for purchase. Each credit generates four image variations or one original image from a text prompt.

The beta release included updated content policy enforcement. OpenAI stated it had improved automated and human monitoring systems to reject prompts attempting to generate violent, adult, or political content. The company also introduced visible watermarks on generated images and deployed new detection classifiers to identify whether an image was created by DALL·E.

Access remained restricted despite the beta designation. Users required an OpenAI account and acceptance from the waitlist. The system prohibited uploads of realistic faces and enforced restrictions on generating public figures' likenesses. OpenAI stated these measures aimed to prevent misuse while gathering data on real-world usage patterns.

The announcement did not specify performance benchmarks, error rates, or failure modes users might encounter. No service-level agreements were disclosed. The beta designation indicated the system remained under active development, with behaviour subject to change as OpenAI collected feedback from the expanded user base.

Why this is an AI incident

Launch-archive bulk classification (10 May 2026). Source signal originates from a real AI provider, regulator, or model-comparison probe; the harm or behavioural change described would not have occurred without the AI system being deployed in the role described. Editor reviewing the archive may amend the rationale per-wire.

Counterfactual "but-for" test per the Editor's Guide.

Codes M1, F10
Providers OpenAI