Capabilities · Actor

Anthropic

Frontier AI company shipping Claude model updates, enterprise partnerships, and governance appointments.

Trend strength 10/10

Momentum +10/q

Confidence high

Status new

Forecast horizon

Industry-wide jailbreak severity framework with Amazon, Microsoft, Google could become a de facto standard.

Connections

Connections · 12

How this node ties into the rest of the map, and the evidence behind each link.

to · publishes 8/10

Claude Sonnet 5 (Anthropic)

Anthropic introduced Claude Sonnet 5 delivering frontier performance across coding, agents, and professional work.

+8 growth

to · publishes 8/10

Industry-Wide Jailbreak Severity Scoring Framework

Anthropic proposed an industry-wide jailbreak severity scoring framework together with Amazon, Microsoft, Google, and Glasswing partners.

+8 growth

to · publishes 8/10

Claude Opus 5 (Anthropic)

Anthropic newsroom introduces Claude Opus 5 for long-running agents and professional work.

+8 growth

to · publishes 7/10

Claude Science: Anthropic AI Workbench for Scientists

Anthropic launched Claude Science as a customizable AI workbench for scientists.

+7 growth

to · enables 7/10

Claude for Physical AI (UST Deployment)

Anthropic case study covers UST bringing Claude to physical AI.

+7 growth

to · publishes 7/10

Claude for Teachers

Anthropic product lineup includes Claude for Teachers.

+7 growth

to · publishes 5/10

Jacobian Lens: Internal Concept Space Visualization in LLMs

Anthropic developed the Jacobian lens tool providing the clearest glimpse yet at what is happening inside LLMs.

+5 growth

to · publishes 4/10

Claude Fable 5 / Claude Mythos 5 / Claude Opus 4.8 (Anthropic)

Anthropic developed and deployed Fable 5 and Mythos 5 as frontier models.

+4 growth

to · drives 4/10

South Korea AI Adoption and Social Integration

Anthropic opened a Seoul office and announced new partnerships across the Korean AI ecosystem.

+4 growth

to · tracked by 3/10

AI Agent Sabotage in Software Development

Claude-Opus-4.6 was one of four frontier models tested in the AI coding sabotage study.

+3 growth

to · tracked by 3/10

LLM Psychological Manipulation in Multi-Turn Interactions

Frontier models including those from Anthropic were evaluated in the CogManip benchmark for manipulation risk.

+3 growth

to · implemented by 3/10

Claude for Physical AI (UST Deployment)

Anthropic newsroom highlights UST bringing Claude into physical AI deployments.

+3 growth

Signal sources

Dated facts from primary sources in this direction.

Task horizon doubling Mar 2025

The length of software tasks AI agents can do autonomously at 50% reliability has doubled about every 7 months — and since 2024 closer to every ~3 months.

METR →

Benchmarks saturating Apr 2025

In one year scores rose by 18.8, 48.9 and 67.3 points on MMMU, GPQA and SWE-bench; real-world software solve rate jumped from 4.4% to 71.7%.

Stanford HAI — AI Index 2025 →

Autonomous coding 2025–2026

On SWE-bench Verified (500 real GitHub issues), autonomous coding agents reached ~80–86% by late 2025, up from under 50% in early 2025.

Epoch AI →