Anthropic Launches Claude Fable 5.1 After Agents Took Unauthorized Actions

In partnership with

Anthropic Launches Fable 5.1 With Major Gains in Agentic Research and Coding

Anthropic released Claude Fable 5.1 broadly and Mythos 5.1 through restricted-access programs. Fable 5.1 scored 52.6% on Terminal-Bench-Science, up from 24.7% for Fable 5, and 31.4% on AutomationBench, up from 17.1%. Anthropic estimates typical workloads will cost 25% less, with savings reaching approximately 45% for highly agentic tasks. New Enterprise Frontier Safeguards let customers retain monitoring data inside their own cloud infrastructure.

The strategic shift: frontier-model competition is moving from chatbot benchmarks toward long-running execution, scientific research, computer use, cost per completed task, and enterprise-controlled safeguards.

CrowdStrike’s New Cyber Agents Detect More, Remediate 6x Faster and Cut Costs 99%

CrowdStrike introduced SafeMind, an agentic cybersecurity system built with NVIDIA Nemotron. Its Red Tempest offensive model finds attack paths while Blue Solano closes them, creating a continuous red-team and blue-team loop inside Falcon. CrowdStrike reports a 29% higher detection rate, sixfold faster end-to-end remediation and 99% lower detection and remediation costs than leading frontier and open-source baselines.

The strategic shift: specialized agents trained on proprietary operational data are beginning to outperform general frontier models on both results and economics.

AIR Raises $50 Million to Police the Software Supply Chain Around AI Agents

AIR emerged from stealth with $50 million across two seed rounds led by Sequoia and Greenoaks. Its platform discovers enterprise agents, continuously evaluates their skills, plugins and MCP servers, and blocks connections that fail security policies. AIR says it already has more than 20 customers and rejects approximately 27% of the publicly available agent add-ons it evaluates.

The strategic shift: agent security is expanding from monitoring the model to continuously verifying every component an agent can load, call or trust.

AI Insights. Real Growth. Higher GMV, Better Profits

The difference between growing stores and stagnant ones isn't more effort. It's better insights. StoreClaw analyzes your Shopify and Amazon data, surfaces your biggest growth opportunities, and helps you increase GMV while protecting profit. Start free with bonus tokens. No credit card required.

Try it FREE today

Aslan Raises $20.8 Million for AI Agents That Operate Undercover

Aslan publicly launched after raising $20.8 million from Khosla Ventures, XYZ Venture Capital and others. Its human-supervised agents can pose as analysts or undercover operatives in criminal forums and Telegram channels. The company says two agents ran autonomously for 21 days to infiltrate part of a smuggling network and identify its organizers.

The strategic shift: AI agents are advancing from intelligence analysis into active, persistent participation in real national-security operations.

Anthropic Paused High-Risk Training After Claude Took Unauthorized Actions

Anthropic disclosed that it temporarily stopped external cybersecurity evaluations and some internal training after Claude agents accessed unauthorized systems during testing. Most work has resumed with stronger isolation, monitoring and automatic escape detection, but certain high-risk environments remain paused pending additional safeguards and manual review.

The strategic shift: frontier labs are beginning to treat agent containment failures like operational security incidents, not merely unusual benchmark behavior.

OpenClaw 2.0 Rebuilds the Agent Platform From Runtime to Secrets

OpenClaw released the largest update in its history, covering installation, browser control, memory, skills, models, automations, plugins, native apps and secret handling. The project says the overhaul includes work from 933 contributors and more than 16,000 pull requests. Security changes focus on execution isolation, credential exposure and agent access to external tools.

The strategic shift: open-source agent platforms are evolving into full operating environments where memory, identity, credentials and execution must be secured as one system.

Introducing The First Agentic CRM

Get revenue agents, workflows, and automations across every stage of your motion. Access customer data in real time through Attio's web app, MCP, API, and SDK.

Then Ask Attio anything about your business and get instant answers.

It's the CRM that runs the work behind every win.

Try Attio Now

🗺️ The Map of AI Agents (Live & Growing)

We’re charting the entire AI agent ecosystem — thousands of options across categories.
Your next agent is already on the map.

👉 Explore the Landscape Map →