AI News and Blog Articles

Save time and grow your business with new Gemini tools

Google added two features to the Gemini app built for small business owners: a one-tap link to a company's Google Business Profile and a workspace called Business notebooks. Once connected, Gemini reads the reviews, customer questions, and performance data tied to the profile, then drafts review replies, updates hours, and answers questions about how the business is doing. Announced at Google for Brazil 2026 on June 10, both tools start rolling out globally during June, with the EEA and the UK excluded at launch.

PRC-linked influence operations are targeting AI debates in the US

OpenAI said it banned two China-linked covert influence operations that used ChatGPT to generate social media comments, images, and political cartoons aimed at inflaming US public debates over AI data centers and trade tariffs. Both campaigns scored Category One, the lowest tier on OpenAI's Breakout Scale, meaning they stayed on a single platform and reached no genuine audience. The findings, published June 10, 2026, show how state-linked actors are testing AI tools to manufacture political content cheaply, even when the real-world reach stays small.

Investing in multi-agent AI safety research

Google DeepMind and four partners opened a research funding call worth up to $10 million to study how AI agents behave when many of them interact at scale. The money targets safety risks that surface only when independent agents built by different organizations work together, a gap most current safety testing ignores. Academic and independent researchers worldwide can apply by August 8, 2026, with award decisions expected in autumn 2026.

Acquisitions & M&ARead Summary

5 ways to automate Meta's Conversions API tool with Zapier

Zapier published a guide showing five no-code ways to feed customer actions into Meta's Conversions API, the server-side tool that sends marketing signals straight to Meta without relying on a browser pixel. The workflows automatically pass events like purchases, form submissions, CRM stage changes, and new email subscribers from apps such as Stripe, Shopify, HubSpot, Salesforce, Typeform, and Klaviyo into Meta so ad targeting stays current. Setup runs through Meta's Events Manager, where Zapier is offered as a partner integration. For tools without a ready-made connection, Webhooks by Zapier sends custom data instead.

How to automate Claude with Zapier

You can connect Anthropic's Claude to the rest of your business tools through Zapier, the no-code automation platform that links more than 9,000 apps and supports over 30,000 actions. Instead of typing one prompt at a time, you set up automated workflows, called Zaps, where Claude reads, writes, summarizes, or analyzes content the moment something happens in another app. Zapier offers one-click templates for common jobs such as drafting replies, summarizing feedback, and routing approvals, so non-technical teams set them up without code. Claude Opus 4.8 is available inside Zapier today, alongside Anthropic's newer Mythos model class.

How a two-person SEO shop is building an engine to run twelve clients in thirty minutes a month

Adrian Martinez, who runs the Toronto agency Hire Adrian with his wife, is building an AI delivery system to serve roughly twelve SEO and answer engine optimization clients while keeping the team at two people. The goal is to shrink hands-on work from 10 to 15 hours per client each month down to about 30 minutes, without hiring account managers. He pairs Claude for planning and decisions with Zapier MCP, which connects the agent to WordPress, reporting tools, and image generation so plans turn into real production work.

MarketingRead Summary

How Gourmet Ads uses Zapier MCP to turn Salesforce and Atlassian into a weekly growth report

Gourmet Ads, an 18-year-old digital advertising business, wired Salesforce and Atlassian Confluence together through Zapier MCP so an AI assistant produces a weekly growth report on its own. The report bundles six parts and five recommended actions, each scoped to take under 20 minutes, and it has already caught fake traffic, a broken link, and a sales pipeline gap. President Benjamin Christie says the setup lifted his personal output by 30 to 40 percent and turned a multi-day engineering job into something he tests from his desk.

The 17 best AI marketing tools in 2026

Zapier's 2026 roundup picks 17 AI marketing tools and sorts them by the job a marketer needs done, from making visuals to tracking whether a brand shows up in ChatGPT answers. The core message is that most marketers do not need every AI feature on the market. The author, who tested the tools, argues the bigger win comes from matching a few tools to real workflows and connecting them so they work together instead of in silos.

MarketingRead Summary

From data to decisions: how LSEG is scaling trusted AI

London Stock Exchange Group (LSEG), one of the world's largest financial markets infrastructure firms, deployed OpenAI's ChatGPT Enterprise and OpenAI APIs across its global business, giving an initial 4,000 employees access to AI tools inside a secure environment. The shift cut many product release cycles from three to six months down to two weeks, and trimmed some customer delivery from as long as nine months to roughly four weeks. LSEG built governance in from the start, including model evaluation, human review, and strict data controls.

Scale Robot Reinforcement Learning with NVIDIA Isaac Lab on Amazon SageMaker AI

Amazon Web Services published a technical walkthrough showing how robotics teams train robot control software on Amazon SageMaker AI using NVIDIA Isaac Lab, a GPU-accelerated simulation framework. The example teaches a Unitree H1 humanoid robot to walk across rough, computer-generated terrain by coordinating its 19 joints, running 4,096 simulated robots in parallel on 8 GPUs. AWS offers two ways to run the work, a persistent managed cluster called SageMaker HyperPod and on-demand SageMaker Training Jobs, both built from one shared container image.

Claude Fable 5: Available on Google Cloud

Claude Fable 5, Anthropic's latest frontier model, is now generally available on Google Cloud's Agent Platform as of June 9, 2026. Google positions the release as part of an ongoing effort to bring the newest models to its platform with safeguards built in for general business use. The model targets demanding work such as advanced software development, long-running autonomous agents, and detailed document analysis. It joins Claude Opus 4.8 and Claude Sonnet 4.6, which are already available through the same platform.

Hands-free first notice of loss: Using Strands Agents and Amazon Bedrock AgentCore Browser Tool for intelligent claims intake

AWS published a technical walkthrough of a hands-free first notice of loss (FNOL) intake system that automates the opening step of an insurance claim. The design pairs domain-reasoning agents built on the open-source Strands Agents SDK with the Amazon Bedrock AgentCore Browser Tool, which drives a managed Chrome session to fill claim portals on its own. Instead of handing adjusters a pile of raw photos and recordings, the system tags and scores evidence at submission time so a person starts with pre-analyzed context.

Build an agentic incident triage assistant with Amazon Quick and New Relic

AWS published a guide showing engineering teams how to build an AI agent that handles incident triage from a single prompt. The agent uses Amazon Quick to investigate an outage through the New Relic Model Context Protocol (MCP) Server, write a root cause analysis brief with evidence links, and open a tracked task in Asana for follow-up. The goal is to compress the slow evidence-gathering work that site reliability engineers do across separate tools into one automated flow.

How to unlock true ROI in software development - a deep dive into the latest DORA research

Google Cloud's DORA research team says the return on AI coding tools follows a J-curve: a short productivity dip during early adoption, then a climb to long-term gains. The report's central finding is that AI works as an amplifier, so the biggest returns come from strong engineering foundations and team workflows rather than the tools alone. Nearly all surveyed teams already use AI at work, yet financial results differ widely depending on how well an organization is set up to absorb the change.

Report: GKE Inference Gateway delivers up to 92% faster AI responses

Google Cloud reports that its GKE Inference Gateway cut the time an AI model takes to start responding by 92.8% compared with Amazon's Kubernetes service in an independent benchmark. The gateway routes AI requests to the exact server already holding the relevant data in memory, using a method called prefix caching. The same test showed 15.7% higher throughput and 62.6% lower delay between words as they stream out. Snap reported reusing cached data on 75 to 80% of requests in its own production setup.

Fluid, natural voice translation with Gemini 3.5 Live Translate

Google DeepMind released Gemini 3.5 Live Translate, an audio model that turns spoken words into translated speech within a few seconds while keeping the speaker's intonation, pacing, and pitch. It detects more than 70 languages automatically and is rolling out across the Gemini Live API, the Google Translate apps, and Google Meet. In Meet, the upgrade widens supported languages from 5 to more than 70, enabling over 2,000 language combinations inside a single call.

✨

Google Gemini

Fluid, natural voice translation with Gemini 3.5 Live Translate

Google released Gemini 3.5 Live Translate on June 9, 2026, an audio model that turns spoken words into spoken words in another language within a few seconds. It covers more than 70 languages, detects each one automatically, and keeps the original speaker's intonation, pacing and pitch so the output sounds human rather than robotic. The model ships at once across three products: Google AI Studio for developers, Google Meet for business customers, and the Google Translate app for everyday users.

💡

Google DeepMind

Introducing Gemma 4 12B: a unified, encoder-free multimodal model

Google DeepMind released Gemma 4 12B, an open-weight model with 12 billion parameters that handles text, images, and audio in a single system and runs on a laptop with 16GB of memory. The big change is an encoder-free design that feeds raw image and audio inputs straight into the language model instead of routing them through separate encoder models. It ships under the permissive Apache 2.0 license, and the wider Gemma 4 family has passed 150 million downloads.

💡

Google DeepMind

Powering the future of robotics in Europe

Google DeepMind opened its first robotics accelerator and chose 15 early-stage European startups for a three-month, equity-free program. Selected teams get hands-on mentorship from Google DeepMind engineers, access to the company's AI stack and Gemini Robotics models, and eligibility for up to $350,000 in Google Cloud credits. The cohort spans ten countries and ends with a Demo Day graduation in London in September 2026.

How engineers at Nextdoor use Codex to build without limits

OpenAI published a customer story showing how engineers at Nextdoor, a neighborhood network serving more than 105 million people, use the Codex coding agent running on GPT-5.5 to build software and fix hard problems. Cory Dolphin, Head of Engineering, says one engineer now owns a feature from idea to production, work once split across three teams. According to Nextdoor, the main constraint is no longer engineering effort but deciding what to build next.

What Codex unlocks for Notion

Notion built its AI Voice Input feature with a single engineer in about 3 to 4 hours using OpenAI's Codex coding agent, according to a new OpenAI customer story. Ryan Nystrom, who leads AI Product Engineering at Notion, pointed Codex at an existing mobile version of the feature and had the agent recreate it for Notion's web and desktop apps in close to one pass. The case study positions Codex as a way for small teams to ship production features at the pace of much larger ones.

Detecting and containing AI-powered threats with Google Security Operations agents

Google has detailed how its Security Operations platform works with Google AI Threat Defense to detect, investigate, and contain attacks automatically, including threats from software a company does not control or cannot patch. The system relies on a set of AI agents that write detection rules, triage alerts, and hunt for hidden threats at machine speed. One agent has reduced a typical 30-minute alert review to 60 seconds and has worked through more than 5 million alerts.

🔗

n8n Blog

Production AI Playbook: Complex Agent Patterns

A new engineering guide from n8n argues that production AI agents break down from missing architecture, not from inherent complexity. It lays out repeatable patterns for the moment a single working agent grows into a tangled multi-agent system that nobody can debug. The four focus areas are orchestrator-to-specialist delegation, sub-workflow composition, memory management across sessions, and structured failure handling, each aimed at keeping growing systems modular, testable, and affordable.

The best Docusign alternatives in 2026

More than 20 Docusign alternatives compete in 2026, ranging from free signing apps like SignWell to enterprise tools like Adobe Acrobat Sign and open-source platforms such as DocuSeal. Because all of these methods produce legally binding signatures under US and EU law, the right pick comes down to your signing volume, budget, and the software your team already uses. Many tools offer free tiers covering three to five documents a month, while pay-as-you-go options charge per document instead of a flat subscription.

Make integrations: Capabilities, limitations, and when to use Zapier

Make, formerly known as Integromat, is a visual no-code automation platform that connects roughly 3,000 apps and lets business users build multi-step workflows by dragging and connecting modules on a canvas. The Zapier article frames Make as a strong fit for people who want fine-grained visual control over how data moves, while flagging two trade-offs: a steeper learning curve and operations-based pricing that adds up as workflows grow. By contrast, Zapier supports more than 9,000 apps and aims for faster setup with less ongoing upkeep. The short version is that the right choice depends on whether you value detailed control or speed and broad app coverage.

The 5 best workflow orchestration tools in 2026

Zapier's 2026 guide ranks five workflow orchestration tools for businesses that need to coordinate multi-step processes across many apps, not run single one-off automations. The five are Zapier (best for building safely with AI, 9,000+ connected apps), Make (visual logic, 3,000+ apps), Workato (enterprise iPaaS, 1,200+ connectors), n8n (self-hosting and code), and Microsoft Power Automate (Microsoft 365 users). Orchestration differs from basic automation because it manages task order, dependencies, retries, and recovery when a step fails.

TidyCal vs. Calendly: Which meeting scheduler is best? [2026]

TidyCal and Calendly both let people book time on your calendar without email tag, but they differ most on cost and depth. TidyCal sells a one-time lifetime license starting at $29, while Calendly charges a recurring monthly subscription and adds far more automation, integrations, and team tools. Cost-focused solo operators and small service businesses tend to fit TidyCal, while larger teams that want analytics, AI meeting notes, and advanced routing fit Calendly.

Regulation & PolicyRead Summary

The best customer experience software in 2026

Zapier's 2026 guide to the best customer experience (CX) software names seven top platforms: HubSpot, Zendesk, Freshdesk, Salesforce, Qualtrics, Zoho, and Intercom. The common thread across every pick is built-in AI, including autonomous agents that resolve support tickets, copilots that draft replies for human staff, and analytics that flag customer frustration before people churn. The guide defines CX software as a way to understand the full customer journey, every touchpoint from first marketing contact through ongoing support, rather than a single moment of sale.

Claude Fable 5 and Claude Mythos 5

Anthropic launched two flagship AI models on June 9, 2026: Claude Fable 5, made safe for general public use, and Claude Mythos 5, a restricted version released first to cyberdefenders. Both share the same underlying model, the company's most capable to date, and both are priced at $10 per million input tokens and $50 per million output tokens, less than half the price of the earlier Claude Mythos Preview. Fable 5 ships with safeguards that route sensitive queries to the older Claude Opus 4.8 model, while Mythos 5 has those safeguards lifted in some areas for trusted partners.

Industrial policy for the Intelligence Age

OpenAI released a roughly 13-page policy blueprint, "Industrial Policy for the Intelligence Age," that asks governments to rewrite taxes, benefits, and infrastructure as advanced AI reshapes work. The headline ideas are a public wealth fund giving every citizen a financial stake in AI growth, a robot tax on companies that replace workers with automation, and a subsidized four-day, 32-hour workweek. OpenAI frames the document as a starting point for debate rather than fixed policy and is funding outside research to test the ideas.

🔗

n8n Blog

Integration Patterns: How To Choose for Your Architecture

An n8n engineering guide lays out about a dozen integration patterns for moving data and events between business systems, split into data patterns and enterprise patterns. The central point is that real production systems almost always run two or three patterns at once, so choosing them deliberately up front saves a costly rebuild later. The guide offers five plain questions to narrow the choice: direction, latency, scale, volume, and cadence.

LLM Observability: What To Instrument and How To Act on It

LLM observability is the practice of capturing what an AI model saw, why it decided, and what it returned, so silent failures become debuggable. Unlike traditional software where errors are obvious, language models can run perfectly while producing wrong or made-up answers, which makes a visible trail of each decision the only reliable way to catch problems. The approach pairs traces, metrics, and logs with feedback loops so teams fix issues where their AI agents actually run.

Unlocking AI flexibility in Europe: A guide to cross-region inference for EU data processing and model access

Amazon Web Services published a guide explaining how cross-Region Inference (CRIS) on Amazon Bedrock lets European customers run generative AI workloads with more capacity while keeping data inside the European Union. The feature automatically routes model requests across several EU AWS Regions, such as Frankfurt, Ireland, Paris, and Stockholm, so a request starting in Europe is only ever processed in Europe. Traffic stays on the encrypted AWS network and never crosses the public internet, which helps businesses meet data residency and GDPR obligations. For some models the EU cross-Region option also costs less than calling a single Region directly.

It's safe to close your laptop now: Hosting coding agents on Amazon Bedrock AgentCore

Amazon Bedrock AgentCore Runtime gives each agent session its own isolated microVM with a persistent workspace, secure tool access through Gateway, and built-in observability-so you can run Claude Code, Codex, Kiro, and Cursor in parallel without sharing secrets, ports, or filesystems. Close the lid, go to dinner, and pick up where you left off tomorrow.

End-to-end encrypted ML inference with Amazon SageMaker AI and FHE

This blog has previously discussed FHE for ML inference in the post Enable fully homomorphic encryption with Amazon SageMaker endpoints for secure, real-time inferencing, but this post goes a little further. That previous post showed how to implement FHE-based inference 'from scratch' by hand-crafting a linear-regression algorithm using a low-level library called SEAL. Instead, this post shows a much more flexible and higher-level approach based on concrete-ml, a high-level library built specifically for FHE-based inference. It supports several common types of models 'out of the box' and is ev

Amazon Quick ARNs: Cross-account migration and namespace permissions

In this post, we cover the structure of Amazon Quick ARNs and provide a practical mental model for working with them. By the end, you can look at an ARN and immediately understand what it means for your migration strategy, diagnose permission issues faster, and design multi-tenant architectures with confidence.

Modernizing Healthcare: How Alcidion achieved greater stability and performance with AlloyDB

Alcidion, a global provider of smart health solutions, migrated its Miya Precision care platform from a legacy Microsoft SQL Server environment to Google Cloud's AlloyDB for PostgreSQL. The move targeted performance bottlenecks, operational overhead, and stability concerns. Using Google's Database Migration Service and custom synchronization tools, the team completed the cutover in about 15 minutes. The result was faster data processing and a step-change in reliability.

Evaluate your Amazon Nova Sonic voice agent at scale, no microphone required

In this post, we walk you through the Nova Sonic Test Harness, an open source framework that we built to solve both problems. It serves as a rapid iteration tool for tuning system prompts and tool configurations (run a conversation, see results, adjust, repeat) and as a comprehensive evaluation framework for validating voice agent quality at scale. It runs complete multi-turn conversations with Amazon Nova Sonic automatically, evaluates them using LLM-as-judge techniques, and can even detect cases where the model's audio output doesn't match its text output (audio hallucinations). No microphon

Calendly vs. Acuity: Which scheduling app should you use? [2026]

This Zapier comparison tests Calendly against Acuity to help readers pick a scheduling app. Both platforms automate booking, sync with calendars and accept payments, but they target different users. Calendly is positioned as a minimalist meeting scheduler best for virtual meetings, while Acuity is positioned as an appointment booking app best for service-based businesses that take payments.

UiPath pricing: Exploring RPA pricing models

This Zapier guide explains how UiPath's robotic process automation pricing works, arguing that the structure is mostly hidden and hard to budget for without a sales call. It walks through UiPath's two main products, Automation Cloud and Test Cloud, their tier structures, and the two licensing models, Unified pricing and Flex. The piece is written to help readers decide whether UiPath is the right choice and how its pricing compares.

Jun 5, 2026

The 6 best electronic signature apps to sign documents online in 2026

This Zapier guide recommends the best electronic signature apps for signing documents online in 2026. The author, a freelance tech writer based in Ireland who has signed dozens of contracts over a decade-plus, tested more than 40 eSignature apps and selected six. A key finding is that legally binding electronic signatures are largely a solved problem, so switching apps is rarely worth it if your current tool works.

Track Stripe payments to Facebook Conversions events with AI

This Zapier tutorial walks through building an automation that connects Stripe payment data to Facebook Conversions using AI. The workflow triggers on a new Stripe payment, uses AI by Zapier to extract and reformat customer details into the format Facebook Conversions expects, then sends that data to Meta. The goal is to let Meta ad campaigns optimize against real revenue rather than surface metrics like click-through rate.

What's new for Managed Service for Apache Spark clusters

Google Cloud detailed updates to its Managed Service for Apache Spark, the rebranded Dataproc service, focused on the managed clusters deployment mode. The improvements center on three pillars: making Spark faster, easier to run, and smarter with embedded AI. Announced at Google Cloud Next '26, the headline addition is the Lightning Engine native execution engine, alongside Flexible VMs and FinOps features for cost control.

MarketingRead Summary

Jun 3, 2026

Automate model quota request and operational issue triage on Amazon Bedrock

In this post, we introduce Amazon Bedrock Ops Alert, a three-layer automated monitoring solution that proactively detects operational issues, dynamically adjusts alarm thresholds, classifies alarms by category, automatically creates context-aware support cases, helps prevent duplicate cases when an unresolved case of the same alarm category is already active, and delivers contextualized notifications to AI SRE teams. We walk through the solution architecture and how you can deploy it in your own environment.

Jun 1, 2026

The fully-managed Remote MCP Server for AlloyDB is now Generally Available

Google Cloud announced that the fully-managed Remote Model Context Protocol (MCP) Server for AlloyDB is now generally available. MCP is an open-source standard that gives large language models a secure, consistent way to connect to external data sources, and this server lets AI agents securely access enterprise data in AlloyDB. The remote server runs on managed Google Cloud infrastructure and exposes an HTTP endpoint for connecting AI applications to data.

Jun 1, 2026

How Trustpilot built a real-time architecture for data enrichment using Gemma

Google Cloud describes how Trustpilot built a real-time architecture to enrich millions of user reviews using fine-tuned Gemma models. Rather than relying on a closed frontier model, Trustpilot fine-tuned open-weight google/gemma-2-9b into a suite of specialized models for tasks like named entity recognition, topic classification, and sentiment extraction. The streaming pipeline runs on Dataflow and Agent Platform endpoints, decoupling business logic from raw LLM inference.

Cool stuff Google Cloud customers built, May edition: Agentic algorithms for supply chains; virtual try-on APIs; robotic camera operators & more

Google Cloud's monthly round-up highlights real-world projects its customers built using AI and cloud technology. The May edition features Urban Outfitters migrating its order management database, BASF mapping global supply chains with AlphaEvolve, and other customers using virtual try-on APIs and robotic camera operators. Each example shows how organizations applied Google Cloud tools to improve performance, cost or capability.

Regulation & PolicyRead Summary

May 29, 2026

Developer's guide to Gemini Enterprise and A2UI integration

This Google Cloud developer guide explains A2UI, an open protocol for agent-driven user interfaces, and how to integrate an A2UI-enabled agent with Gemini Enterprise. A2UI lets an agent return a JSON payload describing a user interface rather than only text or HTML, so it can render rich, interactive UI such as a date picker or map inside a chat surface. The guide uses a restaurant-finder agent built with the Google Agent Development Kit, the A2A protocol, and Gemini as its reference.

Cloud CISO Perspectives: How to build an AI-ready security program for the public sector

In this Cloud CISO Perspectives newsletter, Usman Chaudhary, Field CISO for Google Public Sector, lays out how security leaders at government agencies and critical infrastructure operators can build an AI-ready security program. The guide stresses moving past reactive measures, reducing administrative toil first, and then shifting toward posture elevation, proactive hunting, and structural integration over the next six to 12 months. It describes a mix of building custom internal workflows, buying commercial AI capabilities, and integrating them into an existing security stack.

May 29, 2026

Copilot Health: Now in Preview

Microsoft announced that Copilot Health, first announced in March, has moved into preview after safety testing and evaluation. The product brings a person's health questions, wearable data, and health records together in a secure space within Copilot to provide personalized insights and guidance. It is available starting now to Copilot users in the US aged 18 and over with a Microsoft 365 Personal, Family, or Premium subscription.

Evolving Dataflow to process massive datasets for machine learning

Google created MapReduce more than 20 years ago to handle its early data-processing scaling problems. The company has since evolved its internal data platform, Flume, the successor to MapReduce, with work focused on scalability, efficiency, and developer experience. Many of those features are now available in Dataflow, Google's fully managed batch and streaming platform. The post explains the new capabilities and how Google Cloud customers apply them.

Nano Banana 2 and Nano Banana Pro are generally available, and already powering creative workflows

Google announced that Nano Banana 2 (Gemini 3.1 Flash Image) and Nano Banana Pro (Gemini 3 Pro Image) are generally available via the Gemini Enterprise Agent Platform. The models add image generation and editing to enterprise creative and agentic workflows. A new preview capability lets Nano Banana 2 accept video files as an input prompt. Partners including Adobe, WPP, and Shopify are already using the models in their products.

May 28, 2026

AI in SRE: Where and how Google is deploying agentic AI to improve operations

Google describes how its Site Reliability Engineering teams are adopting agentic AI, an approach it calls SRE AI, to keep services like Search, Gmail, Maps, YouTube, and Google Cloud reliable. The post explains that AI has driven step-changes in system complexity while also offering new ways to improve the software development lifecycle. Google says it aims to use AI as a force multiplier while maintaining human control, spanning areas from reliability design and documentation to anomaly detection and root cause analysis.

Announcing the newest cohort of the Google for Startups Accelerator: Middle East, North Africa & Turkey

Google announced the newest cohort of its Google for Startups Accelerator for the Middle East, North Africa, and Turkiye (MENA-T). The 15 selected AI-first companies begin a three-month program on June 1. The cohort follows a sixth group that concluded in November 2025, in which 14 startups from 8 countries received more than 230 hours of mentorship. The 2026 program adds resources to help founders navigate geopolitical uncertainty in the region.

AI StartupsRead Summary

May 27, 2026

Introducing Google AI Threat Defense to help you outpace the adversary

Google is launching Google AI Threat Defense, an automated security system meant to continuously monitor for and stop AI-powered threats before they affect a business. The company says AI has changed the threat landscape, letting attackers find security gaps faster than teams can fix them manually. The system combines Gemini and other frontier models with the risk prioritization of Wiz, the code remediation of Gemini and CodeMender, and the frontline expertise of Mandiant. It follows a four-step framework of prepare, scan and prioritize, remediate, and monitor.

Acquisitions & M&ARead Summary

May 22, 2026

The Blueprint: How Movix fills a gap in dental skills with specialized agentic AI

Movix is building one of the first agentic AI solutions for dental appliance manufacturers and dental labs, aimed at the shortage of skilled dental technicians. The company started in 2025 and uses Google Cloud infrastructure to detect defects in intraoral scans and automate quality control. Its goal is to help labs scale clinical workflows and reduce costly errors, such as remaking an aligner. The article is part of Google Cloud's Blueprint feature on customers using its AI and cloud technology.

May 21, 2026

How Glance turns hours of video into mobile-ready clips with AI

Google Cloud describes how Glance, a mobile-first content platform, uses AI to turn 1-2 hour horizontal videos into short vertical clips for mobile lock screens. With daily volume projected to grow from 3,500 to over 10,000 videos per day, manual editing was not viable, so Glance built an automated pipeline. The solution combines Google Cloud Speech-to-Text v2, Gemini and the Google Vision API with open-source tools to identify key moments, detect active speakers and reframe video.

Our response to the TanStack npm supply chain attack

OpenAI details its response to the TanStack "Mini Shai-Hulud" supply chain attack, outlines protections taken to secure systems and signing certificates, and explains why macOS users must update OpenAI apps by June 12, 2026. Learn what happened, what was affected, and how OpenAI is strengthening defenses against evolving software supply chain threats.

Feb 26, 2026

Copilot Tasks: From Answers to Actions

Microsoft introduced Copilot Tasks, an AI feature designed to complete work rather than only chat, marking what the company calls the shift from answers to actions. Tasks works in the background with its own computer and browser, across apps and services, and reports back when done. Microsoft launched it as a research preview to a small group of users, with tasks that can run once, on a schedule, or recurring, and built-in consent steps before meaningful actions.

Main Character Energy: 2025 trend recap

Microsoft's Copilot team published a lighthearted 2025 trend recap from MAI, summarizing what people talked about with Copilot over the year. The post frames Copilot's mission as making the user the main character and shares playful comparisons of trending topics, with stats curated by MAI technical staff member Sophia Chen. It points readers to a fuller MAI blog post and paper for deeper analysis of how people use Copilot over time.

Evaluating chain-of-thought monitorability

OpenAI introduces a new framework and evaluation suite for chain-of-thought monitorability, covering 13 evaluations across 24 environments. Our findings show that monitoring a model's internal reasoning is far more effective than monitoring outputs alone, offering a promising path toward scalable control as AI systems grow more capable.

Dec 18, 2025

Deepening our collaboration with the U.S. Department of Energy

OpenAI and the U.S. Department of Energy have signed a memorandum of understanding to deepen collaboration on AI and advanced computing in support of scientific discovery. The agreement builds on ongoing work with national laboratories and helps establish a framework for applying AI to high-impact research across the DOE ecosystem.

Updating our Model Spec with teen protections

OpenAI is updating its Model Spec with new Under-18 Principles that define how ChatGPT should support teens with safe, age-appropriate guidance grounded in developmental science. The update strengthens guardrails, clarifies expected model behavior in higher-risk situations, and builds on our broader work to improve teen safety across ChatGPT.

Dec 17, 2025

Introducing OpenAI Academy for News Organizations

OpenAI is launching the OpenAI Academy for News Organizations, a new learning hub built with the American Journalism Project and The Lenfest Institute to help newsrooms use AI effectively. The Academy offers training, practical use cases, and responsible-use guidance to support journalists, editors, and publishers as they adopt AI in their reporting and operations.

BBVA and OpenAI collaborate to transform global banking

BBVA is expanding its work with OpenAI through a multi-year AI transformation program, rolling out ChatGPT Enterprise to all 120,000 employees. Together, the companies will develop AI solutions that enhance customer interactions, streamline operations, and help build an AI-native banking experience.

Advancing science and math with GPT-5.2

GPT-5.2 is OpenAI's strongest model yet for math and science, setting new state-of-the-art results on benchmarks like GPQA Diamond and FrontierMath. This post shows how those gains translate into real research progress, including solving an open theoretical problem and generating reliable mathematical proofs.

Update to GPT-5 System Card: GPT-5.2

GPT-5.2 is the latest model family in the GPT-5 series. The comprehensive safety mitigation approach for these models is largely the same as that described in the GPT-5 System Card and GPT-5.1 System Card. Like OpenAI's other models, the GPT-5.2 models were trained on diverse datasets, including information that is publicly available on the internet, information that we partner with third parties to access, and information that our users or human trainers and researchers provide or generate.

Copilot is leaving WhatsApp and other messaging apps: What's next

Microsoft announced that Copilot will no longer be available on WhatsApp starting January 15, 2026, due to recent updates to WhatsApp's platform policies that prohibit all LLM chatbots. Copilot will also stop working on other third-party messaging apps on the same date. Microsoft is directing users to continue using Copilot through its mobile app, website, and Windows and Mac desktop apps.

OpenAI and Foxconn collaborate to strengthen U.S. manufacturing across the AI supply chain

OpenAI and Foxconn are collaborating to design and manufacture next-generation AI infrastructure hardware in the U.S. The partnership will develop multiple generations of data-center systems, strengthen U.S. supply chains, and build key components domestically to accelerate advanced AI infrastructure.

Bringing the best of AI search to Copilot

Microsoft announced it is bringing AI search features to Copilot, combining generated answers with cited sources. Starting with the Fall Release, Copilot responses include more prominent, clickable citations and an option to see aggregated sources. The company is also launching a dedicated search experience inside Copilot for greater depth and precision. The update builds on Copilot Search, which began inside Bing.

gpt-oss-safeguard technical report

gpt-oss-safeguard-120b and gpt-oss-safeguard-20b are two open-weight reasoning models post-trained from the gpt-oss models and trained to reason from a provided policy in order to label content under that policy. In this report, we describe gpt-oss-safeguard's capabilities and provide our baseline safety evaluations on the gpt-oss-safeguard models, using the underlying gpt-oss models as a baseline. For more information about the development and architecture of the underlying gpt-oss models, see the original gpt-oss model model card⁠.

💡

Google DeepMind

Oct 24, 2025

Advanced version of Gemini with Deep Think officially achieves gold-medal standard at the International Mathematical Olympiad

Google DeepMind announced that an advanced version of Gemini with Deep Think achieved gold-medal standard at the International Mathematical Olympiad 2025, solving five of six problems for 35 points. Unlike the prior year's silver-medal result, this model operated end-to-end in natural language within the 4.5-hour competition time limit. The results were officially graded and certified by IMO coordinators using the same criteria as for student solutions.

Oct 23, 2025

Human-centered AI

Microsoft AI announced the Copilot Fall Release, which it describes as a step toward AI that is more personal, useful and human-centered. The post frames Copilot as an AI companion that helps people think, plan and connect on their own terms rather than chasing engagement or screen time. The release delivers 12 new features, including shared Groups chats for up to 32 people and a collaborative space called Imagine, along with an on-screen character named Mico.

Sep 30, 2025

Sora 2 System Card

Sora 2 is our new state of the art video and audio generation model. Building on the foundation of Sora, this new model introduces capabilities that have been difficult for prior video models to achieve- such as more accurate physics, sharper realism, synchronized audio, enhanced steerability, and an expanded stylistic range.

Sep 29, 2025

Introducing a more immersive chat experience with Copilot Portraits

Microsoft introduced Copilot Portraits, a set of animated portraits that add visual presence to the Copilot Voice experience. The 40 pre-selected, stylized portraits include real-time voice and animations during conversations to provide a more natural flow than text-based interactions. Released through Copilot Labs, Portraits is available to a limited set of users in Canada, the UK, and the US, with safety measures including session and daily time limits and an 18-and-over age requirement.

Addendum to GPT-5 system card: GPT-5-Codex

This addendum to the GPT-5 system card shares a new model: GPT-5-Codex, a version of GPT-5 further optimized for agentic coding in Codex. GPT-5-Codex adjusts its thinking effort more dynamically based on task complexity, responding quickly to simple conversational queries or small tasks, while independently working for longer on more complex tasks.

A smarter way to talk to your TV: Microsoft Copilot launches on Samsung TVs and monitors

Microsoft announced the launch of Copilot on select Samsung TVs and monitors, bringing its voice-powered AI to the biggest screen in the home at no cost. Copilot offers a voice interface, a friendly animated on-screen character, and visual cards for tasks like recaps, recommendations, and everyday questions. The feature is positioned as a shared, social experience for families and groups.

Release notes: August 7, 2025

Microsoft's Copilot Release Notes for August 7, 2025 announce that GPT-5 is now live in Microsoft Copilot across all markets and platforms, including web, Windows, Mac, and mobile. The same update introduces Smart Mode, which automatically routes users to the most capable model, along with Copilot Mode in Microsoft Edge and Copilot Vision integration in moto ai on select Motorola devices.

Open Source AIRead Summary

Aug 7, 2025

Microsoft incorporates OpenAI's GPT-5 into consumer, developer and enterprise offerings

Microsoft announced it is incorporating OpenAI's GPT-5 across its consumer, developer, and enterprise products to add new reasoning capabilities and improvements to coding and chat. GPT-5, which was trained on Azure, includes OpenAI's latest reasoning models plus a smaller efficient model, with a real-time router that picks the right model for each task. The rollout spans Microsoft 365 Copilot, Microsoft Copilot, GitHub Copilot and Visual Studio Code, and Azure AI Foundry, starting the same day.

Open Weights and AI for All

AI's next frontier isn't just about capability-it's about who gets to use it. Our mission to put AI in the hands of as many people as possible is what drives us. Today's release of our most capable open-weights models is a major step forward that makes advanced AI more open, flexible, and accessible worldwide.

Aug 5, 2025

Introducing gpt-oss

We're releasing gpt-oss-120b and gpt-oss-20b-two state-of-the-art open-weight language models that deliver strong real-world performance at low cost. Available under the flexible Apache 2.0 license, these models outperform similarly sized open models on reasoning tasks, demonstrate strong tool use capabilities, and are optimized for efficient deployment on consumer hardware.

Jul 22, 2025

Stargate advances with 4.5 GW partnership with Oracle

Oracle and OpenAI have entered an agreement to develop 4.5 gigawatts of additional Stargate data center capacity in the U.S. This investment will create new jobs, accelerate America's reindustrialization, and help advance U.S. AI leadership. It also marks a major milestone for Stargate, OpenAI's AI infrastructure platform and long-term vision to deliver the benefits of AI to everyone.

Jul 21, 2025

AI as the greatest source of empowerment for all

I've always considered myself a pragmatic technologist-someone who loves technology not for its own sake, but for the direct impact it can have on people's lives. That's what makes this job so exciting, since I believe AI will unlock more opportunities for more people than any other technology in history. If we get this right, AI can give everyone more power than ever.

Jul 17, 2025

OpenAI nonprofit jam

At OpenAI, we build tools to help people solve hard problems-including nonprofits working on the frontlines of their communities. The OpenAI Academy is teaming up with the Walton Family Foundation, Emerson Collective, and a network of local nonprofit organizations to host the Nonprofit Jam-a one-day, nationwide event bringing together more than 1,000 nonprofit leaders across 10 locations.

AI in Australia-OpenAI's Economic Blueprint

Today, OpenAI, in partnership with Mandala Partners, is sharing the OpenAI AI Economic Blueprint for Australia. At a time when boosting productivity has emerged as a national priority for Australia, the Blueprint provides a clear, actionable plan for how Australia can unlock the full economic and social potential of artificial intelligence.

Addendum to o3 and o4-mini system card: Codex

Codex is a cloud-based coding agent. Codex is powered by codex-1, a version of OpenAI o3 optimized for software engineering. codex-1 was trained using reinforcement learning on real-world coding tasks in a variety of environments to generate code that closely mirrors human style and PR preferences, adheres precisely to instructions, and iteratively runs tests until passing results are achieved.

Introducing AI stories: daily benefits shine a light on bigger opportunities

Sam Altman has written that we are entering the Intelligence Age, a time when AI will help people become dramatically more capable. The biggest problems of today-across science, medicine, education, national defense-will no longer seem intractable, but will in fact be solvable. New horizons of possibility and prosperity will open up.

🧠

Anthropic

Feb 24, 2025

Claude's extended thinking

Lyft to bring Claude to more than 40 million riders and over 1 million drivers

Operator System Card

Drawing from OpenAI's established safety frameworks, this document highlights our multi-layered approach, including model and product mitigations we've implemented to protect against prompt engineering and jailbreaks, protect privacy and security, as well as details our external red teaming efforts, safety evaluations, and ongoing work to further refine these safeguards.

Jan 21, 2025

Stargate Infrastructure

OpenAI, and our strategic partners, are thrilled about our shared vision for the Infrastructure of AGI. We are energized by the challenges we face and are excited by the prospect of partnering with firms across the industrial base to deliver against our ambitious mission. Specifically, we want to connect with firms across the built data center infrastructure landscape, from power and land to construction to equipment, and everything in between.

Jun 18, 2024

Surging developer productivity with custom GPTs

Paf adopted ChatGPT Enterprise across its entire company, with engineers using custom GPTs on a daily basis to speed up routine development tasks. Paf also integrated ChatGPT Enterprise into the grit:lab coding academy (gritlab.ax), training the next generation of software developers using an AI-augmented, systems-architecture mindset from day one. In addition to the wide range of use cases for developers and grit:lab students, 70% of Paf employees actively use ChatGPT Enterprise, spanning business teams like finance, HR, marketing, and customer support.

Using GPT-4o reasoning to transform cancer care

Color Health is working with OpenAI to pioneer a new way of accelerating cancer patients' access to treatment. Their new Cancer Copilot application uses GPT-4o to identify missing diagnostics and create tailored workup plans, enabling healthcare providers to make evidence-based decisions about cancer screening and treatment.

Enhancing news in ChatGPT with The Atlantic

The Atlantic is announcing a strategic content and product partnership with OpenAI, which positions The Atlantic as a premium news source within OpenAI. The Atlantic's articles will be discoverable within OpenAI's products, including ChatGPT, and as a partner, The Atlantic will help to shape how news is surfaced and presented in future real-time discovery products.

Creating an AI-powered Magic Studio

Canva is a visual communication platform, enjoyed by more than 175 million people monthly to make presentations, videos, documents, websites, social media graphics and more. A majority of the world's knowledge workers lack design training, but Canva's combination of an easy-to-use interface, vast libraries, and time-saving tools allows anyone to create visually compelling content.

Feb 15, 2024

Video generation models as world simulators

We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios. We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of high fidelity video. Our results suggest that scaling video generation models is a promising path towards building general purpose simulators of the physical world.

Jan 31, 2024

Building an early warning system for LLM-aided biological threat creation

We're developing a blueprint for evaluating the risk that a large language model (LLM) could aid someone in creating a biological threat. In an evaluation involving both biology experts and students, we found that GPT-4 provides at most a mild uplift in biological threat creation accuracy. While this uplift is not large enough to be conclusive, our finding is a starting point for continued research and community deliberation.

May 31, 2023

Improving mathematical reasoning with process supervision

We've trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning ("process supervision") instead of simply rewarding the correct final answer ("outcome supervision"). In addition to boosting performance relative to outcome supervision, process supervision also has an important alignment benefit: it directly trains the model to produce a chain-of-thought that is endorsed by humans.

AI SafetyRead Summary