Skip to content
AI Panic Room
AI Radar

What's new in AI, in plain English

New AI models and work tools for your job. Found and written up automatically every day from official sites, each with a link back to the original. Not an endorsement.

Showing stories for every job.

UpdateGoogleNew

Gemini adds a small-business hub, saved brand styles and finance tools

Google is adding a business landing page to Gemini, alongside tools for branded visuals and financial analysis. The update also connects Gemini with business apps including Shopify, Canva, Xero and Google Business Profile.

Read the story →
UpdateGoogle

Gemini adds website design, marketing and finance tools for businesses

Google has announced Gemini features for designing business websites, creating on-brand visuals and analysing financial records. New connections include Google business tools and apps such as Canva, Shopify and Xero.

Read the story →
Mistral
New launchMistral

Mistral launches Voice Agents for real-time voice applications

Mistral has released Voice Agents, a managed pipeline for two-way voice interactions in real time. Its SDK offers telephony, browser and custom-audio interfaces.

Read the story →
New launchMistral

Mistral releases Voxtral 3 speech models in public preview

Mistral has released four Voxtral 3 models for speech transcription and speech generation. The family includes real-time transcription, timestamped transcripts with speaker identification, and text-to-speech with voice cloning.

Read the story →
UpdateOpenAI

OpenAI Agents API adds spreadsheet, slide and Word document skills

OpenAI-hosted sandboxes in the Agents API now include tools for creating and editing common office files. The update adds the Artifact SDK and skills for spreadsheets, presentations and Word documents.

Read the story →
UpdateFigma

Figma Make adds Payload for managing website content and data

Figma is bringing Payload’s content-management capabilities into Figma Make. Teams can build web apps, manage their content and data, and host them in Figma.

Read the story →
New launchOpenAI

Astra for Law pairs GPT-6 with a dedicated legal search index

OpenAI introduced Astra for Law, pairing GPT-6 Astra with a legal search index and instructions for legal analysis and writing. Access is limited to lawyers and supervised staff at selected law firms.

Read the story →
UpdateServiceNow

ServiceNow EmployeeWorks adds no-code assistants and background agents

EmployeeWorks now lets employees build customized AI assistants and repeatable background automations without coding. The new features are in controlled availability, with built-in research and scheduling assistants also offered.

Read the story →
New launchMeta

Meta launches Muse Spark Open with downloadable, modifiable weights

Meta has launched Muse Spark Open, an open-weight AI model built for coding and agentic work. Developers can download and modify its parameters, but running it requires significant computing power.

Read the story →
New launchGoogle

Gemini 4 Argon becomes generally available for coding and agents

Google has released Gemini 4 Argon as a generally available model in the Gemini API. It is built for complex coding, agent-based workflows, math and science tasks.

Read the story →
New launchGoogle

Google launches Gemini 4 Argon for complex, multi-step work

Google has released Gemini 4 Argon to developers and businesses worldwide, targeting complex tasks in coding, finance, law and other fields. Access is paid, and launch API rates are set to double after November 30.

Read the story →
New launchDeepSeek

DeepSeek-V4.1-Flash launches with native visual understanding

DeepSeek has released V4.1-Flash on its API with built-in visual understanding. It replaces the previous Flash models, and DeepSeek says API prices have been reduced.

Read the story →
New launchMistral

Mistral Large 4 enters public preview with a 1M context window

Mistral Large 4 is now in public preview as a general-purpose multimodal model with a 1M context window. Open weights are promised soon, with a 50% launch discount for two weeks.

Read the story →
New launchxAI

Grok 4.7 arrives with configurable reasoning and 500k-token context

xAI lists Grok 4.7 as its new flagship model for coding, chat and tasks that use tools. It has a 500,000-token context window, with API pricing of $2 per million input tokens and $6 per million output tokens.

Read the story →
New launchAnthropic

Claude Sonnet 5.5 promises faster work at lower cost

Anthropic has introduced Claude Sonnet 5.5 as an upgrade over Sonnet 5. It says the model runs 30% faster and costs up to 30% less for most work.

Read the story →
New launchAnthropic

Claude Haiku 5.5 launches for high-volume, cost-sensitive work

Anthropic has introduced Claude Haiku 5.5, a small model designed for high-volume, cost-sensitive work. The company describes it as its fastest, cheapest and most capable small model yet.

Read the story →
New launchHireVueNew

HireVue launches AI Interviewer with follow-up questions

HireVue announced an AI Interviewer that conducts two-way conversations with job candidates and asks follow-up questions. It is available around the clock and scores qualification and skill answers against defined rubrics.

Read the story →
UpdateThomson ReutersNew

CoCounsel Legal brings AI research into Smokeball practice management

A new Smokeball partnership brings CoCounsel Legal and Westlaw Deep Research into its practice management platform. Small law firms can move from a matter file to AI-assisted research without leaving their workflow.

Read the story →
UpdateGoogleNew

NotebookLM becomes Gemini Notebook, with existing notebooks preserved

Google says NotebookLM is now called Gemini Notebook as of July 2026. Existing notebooks remain accessible, and the product continues as a standalone tool on web and mobile.

Read the story →
New launchPicsartNew

Picsart launches Marketing Studio for Google ad creation and testing

Picsart launched Marketing Studio to generate and test Google ad campaigns. It creates visuals and copy, exports assets directly to Google Ads, and measures performance.

Read the story →
New launchNvidiaNew

Nvidia launches open models for enterprise and PC agents

Nvidia has launched new open models to run AI agents in enterprises and on PCs. The listing does not identify the models or explain their requirements.

Read the story →
New launchRecursiveNew

Recursive launches Flint, a coding model for laptops

Recursive has launched Flint, described as a laptop coding model. The company says it was largely built by AI.

Read the story →
New launchCanvaNew

Canva releases its first image generation model

Canva has released its first image generation model. The listing does not explain how to access it or what it costs.

Read the story →
New launchHugging Face

Hugging Face launches Carbon-X for DNA analysis and generation

Hugging Face is releasing Carbon-X, an open-source model that lets researchers ask plain-language questions about DNA and generate new sequences. Researchers can run it on their own computers without sending sensitive patient data to external AI companies.

Read the story →
UpdateTypeSafe AI

Jev 1.14 adds experimental images, bounding boxes and longer context

TypeSafe AI has released Jev 1.14.0 in preview, with small improvements across the model and the largest expected gain in multilingual support. It also introduced experimental image support, longer context and a new bounding-box output type, with warnings against production use.

Read the story →
New launchAnthropic

Claude API launches an Anthropic-managed option on Google Cloud

Anthropic has launched a Claude API service accessed through your Google Cloud project. Google Cloud manages access permissions and marketplace billing, while Anthropic operates the API.

Read the story →
UpdateOpenAI

OpenAI Decisions API leaves beta and doubles its question limit

The Decisions API is now generally available with GPT-6 Luna. It supports up to 400 questions per request, plus shared-input caching, images in question instructions and direct structured-data input.

Read the story →
UpdateAnthropic

Claude API adds customer profiles and restriction-status tracking

New user profiles let API providers identify which end customer each request belongs to. Added status fields let teams see which customers Anthropic has restricted or banned.

Read the story →
New launchONE Data

Development Finance Observatory launches AI queries for aid and debt data

ONE Data has launched a free AI platform for exploring aid, lending, budgets and debt in developing economies. The Development Finance Observatory brings scattered datasets together and lets users ask questions in natural language.

Read the story →
New launchOpenAI

GPT-6 Sol and Luna target lower-cost everyday work

OpenAI announced GPT-6 Sol and GPT-6 Luna as lower-cost models for everyday professional work. The page highlights high-volume summarization and information extraction as intended uses.

Read the story →
UpdateFigma

Figma MCP adds code layers, Specs access and 20 partner integrations

Figma’s MCP update lets connected AI tools generate code layers, work with Specs and create visual assets. Figma is also launching 20 partner integrations, including OpenCode, Dia, Junie and Meta Muse.

Read the story →
New launchFigma

Figma Specs launches to share project context with teams and AI agents

Figma is launching Specs, a shared home for project context on its canvas. Teams can develop specifications together and send them to coding agents through Figma’s MCP server.

Read the story →
New launchCalifornia Department of Human Resources and Office of Data and Innovation

CalCareers launches AI resume matching for California state jobs

CalCareers has launched a Skills Matching prototype that uses AI to suggest state job classifications based on an uploaded resume. It aims to help applicants find suitable job types without first understanding California’s state hiring terminology.

Read the story →
UpdateLaw Insider

Law Insider contract database becomes available in ChatGPT

Law Insider’s contract database is now available in ChatGPT. The SimpleDocs announcement does not explain how access works or which users can use it.

Read the story →
UpdateAbantuHR

AbantuHR plans English-language Athena for conversational training

AbantuHR will introduce an English-language version of Athena, its conversational AI training agent. It turns company manuals and other existing content into guided lessons and checks understanding through interactive assessments.

Read the story →
New launchWix

Symphony by Wix brings coordinated AI agents to small-business work

Symphony by Wix is a standalone platform that builds a team of specialized AI agents around a business’s goals. It connects with almost 70 digital tools and does not require a Wix website.

Read the story →
New launchDeepSeek

DeepSeek opens its search engine through a Search & Fetch API

DeepSeek has opened access to its in-house search engine through an API for web search and fetching pages. It supports multilingual search and includes filtering for spam and low-quality AI-generated content.

Read the story →
UpdateSullivan & Cromwell, in collaboration with OpenAI

Sullivan & Cromwell expands Agreement Analyzer for multiple lawyers

Sullivan & Cromwell developed Agreement Analyzer from a single-user prototype into an agreement-review tool that multiple lawyers can use at once. It combines firm precedents, negotiating playbooks and ChatGPT Enterprise to produce a memorandum and proposed tracked changes.

Read the story →
UpdateApple

Apple previews rebuilt Siri AI for iOS 27

Apple previewed a new generation of Apple Intelligence and a rebuilt conversational assistant called Siri AI. Both are expected later in 2026 with iOS 27, with hardware and regional restrictions.

Read the story →
New launchMeta

Meta launches an open-weight AI model that runs on Macs

Meta has launched an open-weight AI model that can run on Macs, according to a linked report. The page does not specify supported Mac hardware or describe the model’s capabilities.

Read the story →
UpdateReflection

Reflection Beam introduced on AWS for agentic coding

Reflection Beam is introduced as an open-weight model for agentic coding on AWS. The page also lists Beam 1.0 on OpenRouter, but provides no testing results or setup details.

Read the story →
New launchPrintful

Printful launches AI mockups for online product listings

Printful launched an AI mockup generator for e-commerce workflows. It creates lifestyle product photos, puts print-on-demand designs on virtual models, and exports images to online storefronts.

Read the story →
New launchNova Technologies

Nova Technologies launches AI automation for recurring back-office work

Nova Technologies has launched an AI workflow automation solution for small and mid-sized businesses. It is designed to carry out approved steps for tasks such as matching invoices, moving data and preparing recurring reports, with people reviewing the results.

Read the story →
New launchMeta

Meta releases Muse Spark Open weights for coding and AI agents

Meta is releasing Muse Spark Open for developers to download, run and adapt themselves. The model is designed for coding and AI agents, but Meta is keeping its more advanced technology closed.

Read the story →
New launchGoogle

Gemini Nano Banana 2.1 adds improved image editing and panoramic output

Gemini Nano Banana 2.1 is now generally available for image generation and conversational editing. Google reports improvements in visual quality, instruction following, character consistency and text rendering.

Read the story →
UpdateGoogle

Gemini API automatically redirects older Flash model requests

Google has deprecated Gemini 3.7 Flash and 3.5 Flash and removed them from selection in Google AI Studio. API requests now automatically go to Gemini 3.8 Flash and 3.6 Flash, respectively.

Read the story →
UpdateAnthropic

Claude Opus 5.5 fast mode leaves research preview

Fast mode for Claude Opus 5.5 no longer requires an access request or beta header. It is also now available through Microsoft Foundry for Azure-hosted deployments.

Read the story →
UpdateOpenAI

GPT-Live adds Decisions integration to act during voice conversations

OpenAI added native Decisions integration to GPT-Live. Apps can act while someone is speaking or respond to what the voice model says.

Read the story →
New launchMeta and Sierra

Meta and Sierra announce Personal Agent Protocol for business interactions

Meta and Sierra are developing an open standard for personal AI agents to interact with businesses, with customers controlling access and companies setting permissions. The first specification is planned for later in the announcement month, so this is an early-stage initiative rather than a finished rollout.

Read the story →
New launchWorkday

Workday launches Work Intelligence to map AI’s impact on job tasks

Workday’s new Work Intelligence breaks job roles into tasks and tracks how AI is changing them. It connects skills gaps with hiring-versus-training cost models and personalized learning paths.

Read the story →
UpdateGoogle

Google expands Fairwind access to Argon for cybersecurity defenders

Google is expanding its Fairwind Program to hundreds of additional partners. The program gives vetted cybersecurity defenders a version of Argon without its cyber guardrails.

Read the story →
UpdateDeepSeek

DeepSeek V4 Pro API will remain available after September 14

DeepSeek says it will continue providing V4 Pro API services after September 14, 2026. The billing method will remain unchanged, with further notice promised if anything changes.

Read the story →
New launchnSpire AI

nSpire AI introduces Sage for video practice of difficult conversations

nSpire AI is introducing Sage, a tool for rehearsing high-stakes workplace conversations through video role-play with AI. Employees can create scenarios from their own documents and receive feedback aligned with company methodology.

Read the story →
New launchHiringNode

HiringNode to debut AI screening with human-approved job criteria

HiringNode will debut AI candidate screening that requires human approval of locked job criteria before screening starts. It works on top of an existing applicant tracking system and records each decision in an employer-owned, tamper-evident record.

Read the story →
New launchCascade AI

Cascade AI plans Agent Harness for building no-code HR agents

Cascade AI plans to launch Agent Harness, a no-code interface for creating and managing AI agents. HR and IT administrators can choose workflows and control access by role, department and location.

Read the story →
New launchTypeform

Typeform launches AI Landing Pages with built-in lead capture

Typeform launched AI Landing Pages to create campaign pages and capture leads in one interface. It generates layouts and copy, embeds forms, and personalizes page elements based on where visitors come from.

Read the story →
UpdateMistral

Mistral retires OCR 4.0 and directs users to OCR 4.1

Mistral has deprecated OCR 4.0 and lists September 30, 2026 as its retirement date. Users are directed to OCR 4.1 at the same price.

Read the story →
UpdateAnthropic

Anthropic expands Cyber Verification Program for security professionals

Anthropic is expanding its Cyber Verification Program for qualifying security professionals. The program makes advanced cyber capabilities and reduced blocking classifiers available to those users.

Read the story →
UpdateAnthropic

Claude Max and Team plans now include monthly API credits

Claude Max and Team subscriptions now include monthly API credits. The release notes link to instructions for claiming them, but do not specify the amounts.

Read the story →
UpdateAnthropic

Claude Managed Agents adds background multi-agent workflows

Claude Managed Agents can now write workflows that run multiple agents in phases and combine their results. The beta feature is intended for work with many pieces, such as reviewing hundreds of documents.

Read the story →
UpdateOpenAI

OpenAI API adds an in-product HIPAA support setup flow

Eligible organizations can now enable HIPAA compliance support from their API organization settings. Admins can accept the standard Business Associate Agreement in the same flow.

Read the story →
New launchOpenAI

OpenAI launches Decisions API beta for typed answers

OpenAI released the Decisions API in beta with GPT-6 Luna. The page says it turns text and images into typed answers 10 times faster than the Responses API.

Read the story →
UpdateOpenAI

GPT-6.1 Sol adds Ultrafast mode in the Responses API

GPT-6.1 Sol has a new Ultrafast mode that reduces the time between generated output tokens. It is available to all API users, with US and EU data residency options.

Read the story →
Last check: 10/25/2026, 6:00:03 AMChecked todayStories are written by AI from the official page and only appear when a quote from that page backs them up.Sources and coverage