Full archive

Latest AI News

Search reporting on AI markets, models, products, companies, policy, and culture.

Showing 1-18 of 143Results matching "Agent Operating System"

Today

3 stories

2 hours ago

Experian Taps ServiceNow as First Partner for Its Enterprise AI Agents

The planned connection gives ServiceNow agents a route to Experian’s data, decisioning and governance tools, starting with model-risk management for regulated industries. Experian is taking its AI-agent strategy into enterprise workflows where permissions, audit trails and human review can matter as much as the model. Experian calls the system Agent Operating System.

2 min readRead story

Launch Products

3 hours ago

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. The immediate stakes are operational. The platform processes AI inline, meaning it analyzes frames directly on source files held in Eluvio’s Content Fabric.

3 min readRead story

Funding Startups

10 hours ago

Backbone Raises €4M to Bring AI Automation to Food Compliance Work

The Belgian startup is consolidating quality records now and developing agents to handle routine checks later. Its test is whether that automation can earn trust in factory workflows shaped by specifications, certifications and laboratory results. Pitchdrive led the round.

2 min readRead story

Yesterday

7 stories

Enterprise adoption Business

16 hours ago

GE Appliances Deploys 800 AI Agents, Reports 25% Fewer Back Orders

The deployment puts Gemini Enterprise into manufacturing, warehouse, logistics and supply-chain work. GE Appliances has disclosed a measurable result for one supplier workflow, but not for the full agent network. GE Appliances says it has deployed more than 800 AI agents across manufacturing, warehouses, logistics and supply-chain operations.

2 min readRead story

Launch Models

21 hours ago

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. GPT-6 Astra reached 99.9% on ARC-AGI-3 ’s Semi-Private set at high reasoning effort through a Provider Adapter harness. The benchmark places agents in unfamiliar abstract, turn-based environments without explicit instructions.

3 min readRead story

Launch Models

23 hours ago

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. HUMAIN has put humain-m3, an Arabic-focused AI model built on the MiniMax -M3 lineage, into research preview through HUMAIN Node. Developers can use a no-code playground or an OpenAI-compatible API.

3 min readRead story

Products

23 hours ago

xAI Explains How Grok Bot Is Built to Keep Agents Working Between Chats

xAI’s design sets a Bot—not a disposable conversation—as the durable unit of work. The harder test is whether users can grant agents enough autonomy while retaining clear ways to inspect, redirect, and stop them. xAI is describing Grok Bot as an agent workspace built for work that survives the chat that started it.

4 min readRead story

Launch Models

23 hours ago

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. OpenAI has released GPT-6 Astra to a limited set of customers, starting with participants in its Daybreak program for cybersecurity defenders.

3 min readRead story

Tools

Yesterday

Oracle Publishes Agent Design That Keeps Credentials and Proof Outside the Model

Oracle’s proposed harness treats a model’s output as a request for action, leaving systems outside the model to authorize work, retain state and verify its result. Oracle has published guidance for production AI agents that separates reasoning from operational authority. The distinction addresses a failure that can appear after an agent has done much of the underlying work.

2 min readRead story

Platform shift Business

Yesterday

Sonos 27 Lets ChatGPT Control Music Across Its Speaker System

The Sonos 27 update gives the company a route to make AI a system-wide control layer, while its rollout will test whether Sonos can restore confidence in software after its 2024 app failure. Sonos is adding AI control to its connected-speaker system through Sonos 27, its next-generation audio operating system. Two paths to AI control Sonos 27voice is the company’s own assistant for music and mood curation.

2 min readRead story

Earlier this week

8 stories

Benchmark Infrastructure

Yesterday

InferenceX Puts GB300 First for MiniMax M2.7, but Only on a Fixed Chat Test

The new ranking gives infrastructure teams a comparable speed and cost snapshot, but its single-turn design does not test the multi-step agent work MiniMax highlights for M2.7. MiniMax pitches M2.7 for complex agent harnesses and software-engineering work. GB200 NVL72 came second at 12,515 tokens per second per GPU, a 10% gap.

2 min readRead story

Partnership Tools

Yesterday

LangChain Adds Agent Payments With Spending Caps, but Settlement Questions Remain

The Nevermined integration lets an agent purchase data and API access during a task under delegated-card limits. An open repository issue, however, raises a separate question about whether some paid results can be released when settlement fails. LangChain is offering a path for agents to buy the APIs and data services they need while working, rather than stopping for a person to complete checkout.

3 min readRead story

Launch Business

Yesterday

Meta Drops Token Counts From Reviews While Employees Test Hatch

Meta is separating performance assessment from AI activity just as it puts a more autonomous tool in front of employees. Whether workers trust Hatch with personal accounts remains a separate test. Meta has told employees that performance reviews will no longer depend on how much they use AI tools.

2 min readRead story

Security risk Tools

Yesterday

Capsule Security Adds a Last-Second Block for AI-Agent Actions

The new control layer judges a planned action before it runs, rather than relying only on access permissions or post-incident review. Its reported benchmark results are promising, but the company’s internal model comparison remains difficult to assess fully. Capsule Security has released a runtime control layer that checks an AI agent’s intended action immediately before execution.

2 min readRead story

Launch Products

Yesterday

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. An AI agent’s access can now be treated as a series of decisions rather than a privilege that lasts for a full login session. CrowdStrike highlighted that model, called Continuous Identity for AI Agents, during its Fal.Con 2026 keynote program in Las Vegas.

2 min readRead story

Models

Yesterday

Cohere Urges Enterprises to Match AI Model Size to Each Task

The company’s new guidance treats model selection as an operating decision: use compact systems for bounded work where lower compute and flexible deployment matter, while retaining larger models where the task requires them. Cohere is urging enterprises to stop treating one large model as the default for every AI job. The proposal starts with a broad definition.

2 min readRead story

Funding Startups

Yesterday

HiddenLayer Raises $100M for Runtime Controls on Enterprise AI Agents

The startup is funding a security layer meant to observe and intervene as agents use enterprise tools. Its test is whether buyers will keep purchasing that layer independently of the platforms hosting their AI. HiddenLayer has raised a $100 million Series B to expand security software for enterprise AI systems, with a particular focus on watching AI agents while they operate.

2 min readRead story

Funding Startups

Sep 2, 2026

Empirik Emerges With $21M to Check Infrastructure Changes Before Deployment

The Sequoia-incubated startup wants to calculate a proposed change’s reach before it runs, then block or escalate risky work. Its central test is whether enterprises will trust those assessments in production. Empirik has emerged from stealth with $21 million to build an AI agent that evaluates infrastructure changes before deployment.

2 min readRead story

Load more stories18 more next