Launch: Latest AI News

Eluvio Will Demonstrate AI Video Workflows That Avoid File Copies at IBC

The company’s IBC showcase centers on keeping analysis, metadata and clip creation alongside source media—a design intended to cut transfer and transcoding steps as AI workflows reach live sports and large archives.

Yesterday

3 stories

21 hours agoModels

GPT-6 Astra Hits 99.9% on ARC-AGI-3, but Scores 62.7% in a Shared Test

ARC Prize will label the two evaluation conditions separately after OpenAI’s context-management setup produced a near-perfect result that its shared interface did not reproduce.

23 hours agoModels

HUMAIN Releases Arabic AI Preview Built on MiniMax’s M3 Lineage

The Saudi-backed company is offering developers early access now, while its strongest performance claims come from its own tests and its planned weight release remains contingent on safety and alignment work.

23 hours agoModels

OpenAI Releases GPT-6 Astra With Its First Advanced Cyber Safeguards

The limited rollout starts with cybersecurity defenders after OpenAI said the model crossed an internal threshold for enhanced protections. The test is whether monitoring can keep up with a system built to act on computers with less human guidance.

Earlier this week

14 stories

YesterdayProducts

CrowdStrike Wants AI Agents Rechecked Before Every Action

The company’s approach makes an agent’s authority temporary: access is scoped to a job and can be withdrawn when that job ends or risk conditions change.

Sep 2, 2026Products

Anthropic Gives Retailers Claude Agent Blueprints, With Checkout Still Off-Limits

The new guidance packages shopper recommendations and merchant decision support for a holiday-season retail push, while keeping agents from completing purchases.

Sep 2, 2026Tools

Google Launches Fairwind to Give Selected Defenders AI Tools for Finding and Fixing Flaws

The program draws a line between broadly available code-security tooling and restricted access to Gemini 3.8 Flash Cyber, reflecting the dual-use risk of AI that can identify and repair software weaknesses.

Sep 1, 2026Tools

Perplexity Documents Versioned Agent API Profiles for Pinning Stable Configurations

The configuration layer packages a run’s model, instructions and tools under one ID. Teams can pin a known setup, while requests using latest automatically take newly uploaded versions.

CrowdStrike’s SafeMind Pits Red and Blue AI Agents in an NVIDIA Infrastructure Twin

The new Falcon system combines CrowdStrike’s security data and agent harnesses with NVIDIA’s open Nemotron models. Early tests favor the optimized open pipeline on detection quality, but cover only one attack scenario family.

Altruist Puts Six-Part Financial Planning in Hazel, With Separate Math Programs

The new agent gives advisors a way to build and adjust comprehensive plans during client meetings. Altruist’s key design claim is that AI handles client context while separate software produces the financial figures.

Sep 1, 2026Tools

OpenClaw 2026.8.2 Adds Linux Desktop Access and Safer Agent Session Recovery

The new release moves agent work closer to the desktop and active browser page, while adding recovery checks intended to prevent an update or migration failure from becoming a false success.

Amazon Quick Pitches Connected Workflows as Vertiv Plans 25% User Growth in 2026

AWS is framing Quick as a conversational layer across business systems, but its customer examples describe deployments and intended uses rather than comparable measures of productivity or return.

Alibaba Opens QwenWork Global Beta for Web and Desktop Work, but Leaves Pricing Unset

The new international beta consolidates three Alibaba agent services into one workplace product that can act across browsers and local computers after authorization. Its practical commercial terms remain incomplete: Alibaba has not disclosed international pricing or a full-release date.

Sep 1, 2026Models

Flower Labs Starts Endeavor 1.0 Preview With a Private Deployment Option

The model’s differentiator is not just Flower’s frontier-performance claim. Customers can run it as a managed service or place selected workloads inside their own infrastructure, though the launch remains limited to early users.

Salesforce Pushes Agents Into CRM Workflows as NIST Turns to Identity Controls

The shift is not just from answering questions to taking actions across business systems. It puts permissions, audit trails, change control and traffic capacity ahead of model choice for teams trying to move beyond pilots.

Aug 31, 2026Tools

Google Updates Antigravity Teamwork, Putting Long-Running Agent Critique in Paid Preview

The update packages critique, verification and synthesis into reusable agent-team patterns. Google’s results suggest the orchestration can produce useful work, while the preview’s lower-parallelism setup and model mix limit direct comparisons with its research runs.

Aug 31, 2026Products

Broadcom Releases AgentMinder to Check AI Agent Tool Calls at Runtime

The new control layer is designed to evaluate each requested agent action before it reaches company systems, shifting governance from static permissions toward live policy enforcement.

Aug 31, 2026Products

Replica Cyber Ships Isolated AI Agents With Runtime Controls for Regulated Firms

The product makes containment, intervention and post-incident reconstruction part of the runtime environment. Its unanswered test is whether those controls hold when agents connect to the systems that make them valuable.