Technical Performance · Index
Thank you for reading this post, don't forget to subscribe!
Home / Indexes / Technical Performance / AI Safety & Governance 2026
AI Safety & Governance 2026
Documented AI incidents, the EU AI Act, and the surge in US regulation — the most important primary-sourced AI safety and governance statistics for 2026. Every figure is linked to its primary source and dated.
By The AI Index · Updated July 2026 · 6 min read · 10 sourced figures
Key takeaways
- 362 AI incidents were documented in 2025, up from 233 in 2024 — a ~55% year-over-year rise. (AI Incident Database / Stanford HAI)
- The EU AI Act becomes fully applicable on 2 Aug 2026, with fines up to €35M or 7% of global turnover. (European Commission)
- US states passed 131 AI-related laws in the past year (up from 49 by 2023); federal AI regulations more than doubled to 59 in 2024. (Stanford HAI)
- Governance is maturing unevenly: firms with no responsible-AI policy fell from 24% to 11%, but model transparency declined. (Stanford HAI)
AI incidents and risk
As capability and deployment scale, so do harms. The AI Incident Database logged 362 incidents in 2025, up from 233 in 2024 — a continuation of a steep multi-year climb (Stanford HAI 2026, Responsible AI). Governance is maturing unevenly: businesses with no responsible-AI policy fell from 24% to 11%, AI-governance roles grew +17% in 2025, yet the Foundation Model Transparency Index slipped to 40 from 58 in 2024.
| Year | Documented AI incidents |
|---|---|
| 2024 | 233 |
| 2025 | 362 |
Documented AI incidents, 2024–2025. Source: AI Incident Database / Stanford HAI
The regulatory landscape
Rule-making accelerated sharply. In the US, 59 AI-related federal regulations were introduced in 2024 — more than double the 25 in 2023 — from 42 agencies. At the state level, 131 AI-related laws passed in the past year, up from 49 by 2023 (Stanford HAI, Policy & Governance).
The EU AI Act timeline & penalties
The EU AI Act entered into force on 1 August 2024 and becomes fully applicable on 2 August 2026. Bans on prohibited practices and AI-literacy duties applied from February 2025; rules for general-purpose AI (GPAI) models and governance from August 2025. Finland became the first member state with full enforcement powers in December 2025. Penalties are tiered (European Commission): up to €35M or 7% of turnover for prohibited practices, €15M or 3% for high-risk obligations, and €7.5M or 1% for supplying incorrect information. In August 2025, 26 major AI providers — including Microsoft, Google, Amazon, OpenAI, and Anthropic — signed the GPAI Code of Practice; Meta declined.
The numbers in full
| Metric | Figure | Source |
|---|---|---|
| Documented AI incidents (2024) | 233 | AIID / HAI |
| Documented AI incidents (2025) | 362 | AIID / HAI |
| Businesses with no RAI policy | 24% → 11% | Stanford HAI |
| Growth in AI-governance roles (2025) | +17% | Stanford HAI |
| Foundation Model Transparency Index (2025) | 40 (from 58) | Stanford HAI |
| US federal AI regulations (2024) | 59 (from 25) | Stanford HAI |
| US state AI laws (past year) | 131 (from 49) | Stanford HAI |
| Max EU AI Act fine (prohibited) | €35M / 7% | EU Commission |
Figures reflect each source’s latest release.
Frequently asked
How many AI incidents were reported in 2025?
The AI Incident Database recorded 362 incidents in 2025, up from 233 in 2024, per Stanford HAI’s 2026 AI Index. (AIID / Stanford HAI)
What are the EU AI Act fines?
Up to €35 million or 7% of global annual turnover for prohibited practices; €15M or 3% for high-risk violations; and €7.5M or 1% for supplying incorrect information. (European Commission)
Is AI regulation increasing?
Sharply. US federal AI regulations more than doubled to 59 in 2024, and US states passed 131 AI-related laws in the past year, up from 49 by 2023. (Stanford HAI)
Primary sources
• Stanford HAI — AI Index 2025, Policy & Governance (federal & state AI regulation) — hai.stanford.edu
• European Commission — EU AI Act (timeline · tiered penalties) — digital-strategy.ec.europa.eu