{"event":{"id":"evt-dalle-preview-20220518","dedupe_key":"generative_ai_preview_scaling:2022-05-18:dall-e-2-research-preview-scales-user-interaction-and-deployment-safety-feedback","event_type":"generative_ai_preview_scaling","title":"DALL·E 2 research preview scales user interaction and deployment-safety feedback","summary":"OpenAI reported more than three million generated images from early users, expanded onboarding and changes to filters, automated detection and human review during the DALL·E 2 research preview.","occurred_at":"2022-05-18T00:00:00.000Z","published_at":"2022-05-18T00:00:00.000Z","observed_at":"2026-10-01T18:14:00.000Z","reconstructed_at":"2026-10-01T18:14:00.000Z","ingest_type":"live_world_scan","locations":null,"status":"verified","confidence":0.98,"metadata":{"actors":["OpenAI"],"what_happened_at_time":"A multimodal generative model was being exercised by real users at meaningful preview scale while its safety system was iteratively adjusted from observed use.","evidence_at_time":"OpenAI's May 18 update provided usage counts, expansion plans and specific changes to safety controls; the underlying DALL·E 2 technical paper had appeared in April.","affected_layers":["layer-ai","layer-models","layer-human","layer-software","layer-governance"],"change_kind":"deployment_and_feedback_loop_maturation","q1_continuity":"Q1 already showed humans shaping models through RLHF. Q2 added a deployment loop in which user behavior and violations shaped product-level safety and access controls.","retrospective_inference":"Human-AI interaction was expanding from text/API usage into generative creative workflows, while model deployment itself became an empirical source of governance information.","uncertainty":"The May event was not DALL·E 2's birth: limited preview and technical work predated it. Preview users were selected and therefore not representative of unrestricted public deployment."},"created_at":"2026-10-01T19:22:17.749Z","updated_at":"2026-10-01T19:22:17.749Z"},"artifacts":[{"id":"art-dalle-preview-20220518","canonical_url":"https://openai.com/index/dall-e-2-update/","artifact_type":"research_preview_update","title":"DALL·E 2 research preview expansion","summary":"OpenAI reported more than three million images generated by early DALL·E 2 users, expanded onboarding and iterative changes to filters, automated detection and human review based on preview experience.","creator_entities":["OpenAI"],"released_at":"2022-05-18T00:00:00.000Z","content_hash":null,"metadata":{"historical_scope":"2022-Q2","onset_note":"DALL·E 2 research and limited preview activity began before this update; May 18 is an adoption/deployment-safety event, not the model's birth."},"created_at":"2026-10-01T19:22:17.743Z","updated_at":"2026-10-01T19:22:17.743Z","relation":"documented_by"}],"signals":[{"id":"sig-q2-governance-stack-coupling","statement":"Public and private governance moved closer to the technical stack through AI-risk measurement work, semiconductor/export-control coordination and controlled model deployment/release practices.","signal_type":"governance_maturation","direction":"increasing","confidence":0.95,"epistemic_status":"supported_inference","mapping_mode":"retrospective","reconstructed_at":"2026-10-01T18:14:00.000Z","created_at":"2026-10-01T19:22:17.754Z","updated_at":"2026-10-01T19:22:17.754Z"},{"id":"sig-q2-human-ai-workflow-expansion","statement":"AI systems became more directly embedded in creative and software-development workflows, creating larger real-user feedback surfaces before the later mass-market chatbot phase.","signal_type":"adoption_and_interface_shift","direction":"increasing","confidence":0.97,"epistemic_status":"supported_inference","mapping_mode":"retrospective","reconstructed_at":"2026-10-01T18:14:00.000Z","created_at":"2026-10-01T19:22:17.756Z","updated_at":"2026-10-01T19:22:17.756Z"},{"id":"sig-q2-multimodal-capability","statement":"Model interfaces broadened materially beyond text toward interleaved vision-language understanding and large-scale text-guided image generation in real user workflows.","signal_type":"capability_broadening","direction":"increasing","confidence":0.96,"epistemic_status":"supported_inference","mapping_mode":"retrospective","reconstructed_at":"2026-10-01T18:14:00.000Z","created_at":"2026-10-01T19:22:17.752Z","updated_at":"2026-10-01T19:22:17.752Z"}]}