AsOf

Trends

Safety and policy

Model safety, usage policies, trust and safety, regulation. Counted from the record's checked claims, one column per quarter since 2023. /trends/safety-policy.md

28
2023 Q12026 Q4*
By quarter
2026-Q4*8 of 193 updates (4%)
2026-Q37 of 253 updates (3%)
2026-Q24 of 175 updates (2%)
2026-Q11 of 98 updates (1%)
2025-Q42 of 62 updates (3%)
2025-Q24 of 46 updates (9%)
2025-Q11 of 36 updates (3%)
2024-Q21 of 27 updates (4%)
all quarters2024-Q22025-Q12025-Q22025-Q42026-Q12026-Q22026-Q32026-Q4

1–28 of 28 claims since 2023, newest first

2026-10-28valid
Google · released · Gemini 4 Carbon
Google released: Gemini 4 Carbon, claim clm_0516802A12DE · valid · 1 source (1 primary) · conf 0.70
2026-10-27valid
Cloudflare · launched · PHP Capture the Flag challenge
Cloudflare launched: PHP Capture the Flag challenge, claim clm_50A44CCB7951 · valid · 1 source (1 primary) · conf 0.70
2026-10-27valid
Microsoft · model release · MAI Cyber 1.1 Flash
Microsoft model release: MAI Cyber 1.1 Flash, claim clm_61E141829B00 · valid · 1 source (1 primary) · conf 0.70
2026-10-22valid
Cohere · launched · Confidential Computing on Cohere Model Vault
Cohere launched: Confidential Computing on Cohere Model Vault, claim clm_FB95B6CC521B · valid · 1 source (1 primary) · conf 0.70
2026-10-20valid
Claude · feature · Web fetch domain restrictions
Claude feature: Web fetch domain restrictions, claim clm_FBFD57ACC84D · valid · 1 source (1 primary) · conf 0.70
2026-10-12valid
AI Security Institute · policy · national AI risk defences
AI Security Institute policy: national AI risk defences, claim clm_885C98452084 · valid · 1 source (1 primary) · conf 0.70
2026-10-08valid
Anthropic · launched · Cyber Mission
Anthropic launched: Cyber Mission, claim clm_0C637C075926 · valid · 1 source (1 primary) · conf 0.70
2026-10-06valid
Anthropic · launched · Cyber Verification Program
Anthropic launched: Cyber Verification Program, claim clm_977BA9105325 · valid · 1 source (1 primary) · conf 0.70
2026-09-30valid
Google DeepMind · launched · SynthID Bio
Google DeepMind launched: SynthID Bio, claim clm_79276AF14F01 · valid · 1 source (1 primary) · conf 0.70
2026-09-17valid
Anthropic · launched · Life Sciences Verification Program
Anthropic launched: Life Sciences Verification Program, claim clm_809BA0FBCC61 · valid · 1 source (1 primary) · conf 0.70
2026-09-02valid
Gemini · model release · Gemini 3.8 Flash
Gemini model release: Gemini 3.8 Flash, claim clm_D0AA27CF4B98 · valid · 2 sources (2 primary) · conf 0.70
2026-08-07valid
OpenAI · model release · GPT-5.6 Cyber
OpenAI model release: GPT-5.6 Cyber, claim clm_EF2F941802E7 · valid · 1 source (1 primary) · conf 0.70
2026-08-04valid
Mistral AI · released · Shieldstral
Mistral AI released: Shieldstral, claim clm_E76CBA22820A · valid · 1 source (1 primary) · conf 0.70
2026-07-27valid
Microsoft · launched · Project Perception
Microsoft launched: Project Perception, claim clm_D564025B3CF7 · valid · 1 source (1 primary) · conf 0.70
2026-07-01valid
Anthropic · policy · Restored access to Claude Fable 5
Anthropic policy: Restored access to Claude Fable 5, claim clm_8BB3BD9FE338 · valid · 1 source (1 primary) · conf 0.70
2026-06-23valid
OpenAI · feature · Safety Usage Dashboard
OpenAI feature: Safety Usage Dashboard, claim clm_16A8ADF1515F · valid · 1 source (1 primary) · conf 0.70
2026-06-04unverified
Nvidia · released · Nemotron 3.5 Content Safety
Nvidia released: Nemotron 3.5 Content Safety, claim clm_0B713AD5A235 · new claim · 1 source · conf 0.50
2026-05-29valid
OpenRouter · feature · Guardrails
OpenRouter feature: Guardrails, claim clm_F592D9772E05 · valid · 1 source (1 primary) · conf 0.70
2026-05-05valid
Meta Platforms · feature · AI-powered age assurance
Meta Platforms feature: AI-powered age assurance, claim clm_CA84FB6A31AC · valid · 1 source (1 primary) · conf 0.70
2026-02-17valid
National Institute of Standards and Technology · released · AI Agent Standards Initiative
National Institute of Standards and Technology released: AI Agent Standards Initiative, claim clm_E7DCB015DC73 · valid · 1 source (1 primary) · conf 0.70
2025-12-16valid
Google DeepMind · released · Gemma Scope 2
Google DeepMind released: Gemma Scope 2, claim clm_AF29B75D3733 · valid · 1 source (1 primary) · conf 0.70
2025-10-29valid
Ollama · released · gpt-oss-safeguard
Ollama released: gpt-oss-safeguard, claim clm_943E117AEF88 · valid · 1 source (1 primary) · conf 0.70
2025-06-27valid
Anthropic · launched · Economic Futures Program
Anthropic launched: Economic Futures Program, claim clm_F9FE0C32C237 · valid · 1 source (1 primary) · conf 0.70
2025-06-16valid
Microsoft · launched · Microsoft Sovereign Cloud
Microsoft launched: Microsoft Sovereign Cloud, claim clm_BCF0A726DB62 · valid · 1 source (1 primary) · conf 0.70
2025-06-06valid
Claude · launched · Claude Gov
Claude launched: Claude Gov, claim clm_90FAAB835BB0 · valid · 1 source (1 primary) · conf 0.70
2025-05-22valid
Anthropic · released · Claude Opus 4
Anthropic released: Claude Opus 4, claim clm_FAD5BC475D63 · valid · 1 source (1 primary) · conf 0.70
2025-02-27valid
Anthropic · launched · Transparency Hub
Anthropic launched: Transparency Hub, claim clm_183BB100E343 · valid · 1 source (1 primary) · conf 0.70
2024-05-29valid
Mistral AI · policy · Non-Production License
Mistral AI policy: Non-Production License, claim clm_B4F1154C2222 · valid · 1 source (1 primary) · conf 0.70