数据时间 最近成功读取 不可用
安全
AI safety, security incidents, alignment research, and regulatory developments tracked from public sources with evidence and dates.
Safety metrics
Multi-model comparison of safety indicators (hallucination, jailbreak, bias…)
| 模型 | 幻觉率 |
|---|---|
| Benchmark Reported | 0.091 |
Safety-related changes
Current verified metric updates from public sources
No safety changes yet
Safety-related changes will appear here once qualified public data is available.
Safety channels
Follow AI safety updates from official and topic channels