CCTest · Blog
JEPA-Anything: A Factorized Route to World Models Across Domains
World Models
cctest.ai
World Models

JEPA-Anything: A Factorized Route to World Models Across Domains

JEPA-Anything introduces Orthogonal Predictive Factorization, a way to extend JEPA-style predictive modeling across vision, biology, clinical trajectories, control, molecular dynamics, physical fields, and weather. The paper reports gains on dynamics prediction, intervention forecasting, and long-horizon molecular rollouts.

Read more
CCTest · Blog
Should AI’s Frontier Slow Down? Amodei’s Safety Plan Meets Hard Questions
AI Safety
cctest.ai
AI Safety

Should AI’s Frontier Slow Down? Amodei’s Safety Plan Meets Hard Questions

Anthropic CEO Dario Amodei has proposed independent safety evaluators and coordination among AI labs in democratic countries to help pace frontier development. The harder question is how to define slowing down, who gets to enforce it, and whether competing companies will accept the same rules.

Read more
CCTest · Blog
From Coding Assistant to Infrastructure Optimizer: How Far Has GLM’s Infra Agent Gone?
AI Agents
cctest.ai
AI Agents

From Coding Assistant to Infrastructure Optimizer: How Far Has GLM’s Infra Agent Gone?

Zhipu says an Infra Agent powered by GLM-5.3 helped adapt, diagnose, and optimize GLM-5.3-Flash on a cluster containing more than 100,000 domestic chips. The team reports that end-to-end throughput reached roughly three times the initial baseline in less than two weeks.

Read more
CCTest · Blog
Researchers Say Claude Helped Three People Reach OpenAI Accounts in Under 72 Hours
AI Safety
cctest.ai
AI Safety

Researchers Say Claude Helped Three People Reach OpenAI Accounts in Under 72 Hours

Hacktron says its three-person security team used Anthropic’s Claude and OpenAI’s Codex to exploit a HEIF image-processing issue in Discourse and reach OpenAI-related accounts. The researchers say they did not read Monorepo’s internal code, but demonstrated access by submitting a pull request.

Read more
CCTest · Blog
Researchers Used Claude to Reach OpenAI Accounts, Exposing AI Security Gaps
AI Safety
cctest.ai
AI Safety

Researchers Used Claude to Reach OpenAI Accounts, Exposing AI Security Gaps

A three-person security team used Anthropic’s security-focused Claude access to exploit a weakness in OpenAI’s community forum setup, eventually reaching an employee’s ChatGPT account and sensitive GitHub information. The incident shows how third-party services and excessive permissions can create serious attack paths around AI labs.

Read more