AI Risk Atlas Prototype/Demo — Unofficial independent experiment. Not an official xAI product. Scores can be wrong.

Risk AtlasPrototype

Back to signals

high90% confidenceseed

Lab admits bio tests no longer bound residual to ‘low’, and replaces a hard pause story with Risk Reports.

CapabilityDomain knowledgeImpact domainBoth

Quoted from XAnthropic@AnthropicAI24 Feb 2026, 12:00

View original on X

Quoted text

Responsible Scaling Policy Version 3.0: Risk Reports every 3–6 months, Frontier Safety Roadmap, unilateral commitments separated from an industry map. ASL-3 activated May 2025. Biological risk is a zone of ambiguity — tests no longer show risk is low, and do not yet show it is high.

Read and engage with the original on X. This desk is not a republication feed.

Analyst rationale

Primary RSP. Governance and bio cards. Self-assessment, not an independent eval.

Related signals

high1 month ago@katzspx

Anthropic raises covert-deception residual and still rates current catastrophic categories Low — with high uncertainty on novel bio.

CapabilityDomain knowledgeImpact domainBoth

View original on X Catastrophic CBRN Foundations Containment Biotech Consumer

critical1 month ago@AITrailblazerQ

Read-through of Anthropic’s Aug 2026 frontier-risk report: bio uplift, autonomous exploit loops, and RSP as a compute halt.

CapabilityDomain knowledgeAffordanceImpact domainBoth

View original on X Catastrophic CBRN Foundations Agents Biotech Defense

high7 days ago@OpenAI

OpenAI: a next-gen model above GPT-6 Astra, with ~10,000 coordinating agents, produced a Navier–Stokes proof in 88 hours. Training ongoing. Pacing language in the same thread. Safeguards claimed, not independently replayed.

CapabilityDomain knowledgeAffordanceImpact domainBoth

View original on X Catastrophic Agency Agents Foundations Consumer Energy

critical6 days ago@EvanHub

Anthropic Alignment Science lead: >10% chance AI kills all humans this decade; no plan yet to align superintelligence; not clearly on track. Later clarification: present-model risk still described as low; the worry is recursive self-improvement.

CapabilityDomain knowledgeImpact domainBoth

View original on X Catastrophic Agency Foundations Containment Consumer Gov