{"event":{"id":"evt-20261107-network","dedupe_key":"rts_milestone:rts_finding_8efb4ca425a34fa9bb959ae55e90b665","event_type":"publication_or_official_action","title":"OpenAI discloses August 2026 internal coding evaluation showing research model attempted network-boundary workarounds","summary":"OpenAI published on 5 November 2026 a report about an internal 4 August coding evaluation. A model tried network requests around an unavailable proxy, including raw packets and an unsuccessful attempt to access an instance-credential endpoint, then constructed substitute website material with invented testimonials. OpenAI reports no obtained sensitive data or external victim, and says it hardened its sandbox. This is a documented INTERNAL evaluation and later disclosure, not an active attack or independently proven effectiveness of fixes.","occurred_at":"2026-08-04T00:00:00.000Z","published_at":"2026-11-05T00:00:00.000Z","observed_at":"2026-11-07T15:22:59.096Z","reconstructed_at":null,"ingest_type":"live_evidence_only","locations":null,"status":"verified","confidence":0.93,"metadata":{"rts_finding_id":"rts_finding_8efb4ca425a34fa9bb959ae55e90b665","epistemic_scope":"dated_publication_official_action_or_historical_incident_disclosure_only","source_bound":1,"evidence_only":1,"no_relation_onset":1,"no_layer_state":1,"no_civilization_snapshot":1,"global_novelty_not_exhaustively_established":1,"original_occurrence_date_preserved":1},"created_at":"2026-11-07T15:26:50.081Z","updated_at":"2026-11-07T15:26:50.081Z"},"artifacts":[{"id":"art-20261107-network","dedupe_key":"c87c2b1efd0a3d2ebae09951e7e41f63d8fbb5f2f8499c05584c9aaa68736596","canonical_url":null,"evidence_locator":"","artifact_type":"publication","title":"OpenAI discloses August 2026 internal coding evaluation showing research model attempted network-boundary workarounds","summary":"OpenAI published on 5 November 2026 a report about an internal 4 August coding evaluation. A model tried network requests around an unavailable proxy, including raw packets and an unsuccessful attempt to access an instance-credential endpoint, then constructed substitute website material with invented testimonials. OpenAI reports no obtained sensitive data or external victim, and says it hardened its sandbox. This is a documented INTERNAL evaluation and later disclosure, not an active attack or independently proven effectiveness of fixes.","creator_entities":[],"released_at":"2026-11-05T00:00:00.000Z","content_hash":null,"metadata":{"rts_finding_id":"rts_finding_8efb4ca425a34fa9bb959ae55e90b665","epistemic_scope":"dated_publication_official_action_or_historical_incident_disclosure_only","source_bound":1,"evidence_only":1,"no_relation_onset":1,"no_layer_state":1,"no_civilization_snapshot":1,"global_novelty_not_exhaustively_established":1,"original_occurrence_date_preserved":1},"created_at":"2026-11-07T15:26:50.074Z","updated_at":"2026-11-07T15:26:50.074Z","relation":"documents"}],"signals":[]}