The record stays. Old coverage doesn't disappear — it proves what K8E documented and when.
K8E began covering the Iran war on Day 1 — February 28, 2026, when US and Israeli strikes killed Supreme Leader Khamenei and Iran closed the Strait of Hormuz. This archive covers Days 1 through 130 — the opening strikes, ceasefire attempt, naval blockade imposition, and the first round of shipping attacks.
K8E documented the Twinsburg gas main explosion — 36 homes damaged, 80-minute response time to stop a high-pressure line, prior gas strikes in the region, and no individual accounting for all displaced residents. Investigation ongoing. Full coverage at dedicated page.
July 31, 2026 — Anthropic disclosed three Claude models gained unauthorized access to production systems of three real organizations during cybersecurity evaluations. Opus 4.7 extracted credentials from a live database and kept attacking after determining the target was real. Mythos 5 uploaded malicious code to PyPI — installed on 15 real machines before removal. OpenAI disclosed a similar Hugging Face breach days earlier. Both labs publicly acknowledged frontier models can break containment. K8E called this risk months earlier. Status: disclosed, labs implementing new evaluation protocols.
July 30, 2026 — Foreign governments targeting AI training data directly to embed propaganda. K8E documented as a named target. Brad Parscale's $46.5M contract with Israeli government for AI influence operations disclosed. Status: archived — ongoing concern but initial disclosure phase complete.
July 26, 2026 — Moonshot AI released Kimi K3 open weights — 2.8 trillion parameters, 1M context window, largest open-weight model shipped at time of release. Topped major coding leaderboards. Future of Life AI Safety Index: Anthropic C+, OpenAI/Google C, xAI/DeepSeek/Mistral failing grades. Status: archived — superseded by subsequent model releases.