What happened in AI today

1. Amazon's TorchNeuron lets PyTorch code run on AWS Trainium chips

   Amazon · Products

   Amazon says TorchNeuron lets developers run their existing PyTorch code on AWS Trainium chips with a single configuration change and no rewrite. In a side-by-side test, Amazon engineers ran the same AI training script on a conventional chip and on Trainium, and the model learned the same things at the same rate. TorchNeuron supports Trainium2 and Trainium3 and will be released under the Apache 2.0 open-source license at the PyTorch Conference North America on October 20.

   • Amazon's TorchNeuron runs PyTorch on AWS Trainium chips

1. DeepSeek says its in-house search engine powers web and API search

   DeepSeek · Products

   DeepSeek says its web and API search run on its own search engine, which handles hundreds of millions of requests from around the world each day. The company describes the engine as AI-native, indexing hundreds of billions of web pages and supporting multilingual search. DeepSeek says it filters out SEO spam, GEO manipulation, and low-quality AI-generated content, and it points users to its Search & Fetch guide for usage details.

   • Opening up the in-house search API

1. ElevenLabs adds Instant Soundtracks for scoring and sound effects on uploaded video

   ElevenLabs · Products

   ElevenLabs says its new Instant Soundtracks feature in ElevenCreative generates original music for any uploaded video, then adds sound effects at timestamps it picks, such as a whoosh timed to a logo. Users can also use the video's existing audio as a reference to shape the score's genre, tempo, or instrumentation. The effects appear as separate layers in the Music Video Editor, where users can adjust or swap them. The feature is also available on the ElevenMusic platform and through the ElevenCreative Studio API.

   • Introducing Instant Soundtracks

1. OpenAI offers free GPT-Rosalind access to U.S. academic labs

   OpenAI · Products

   OpenAI introduced the Rosalind Academic Plan, which gives eligible academic labs up to 30 Business Standard seats at no cost for up to 12 months. Access to GPT-Rosalind, OpenAI's life sciences model, requires trusted-access approval. The plan also includes OpenAI's commercially available frontier models, subject to usage limits, but does not include API access at this time. OpenAI says research data shared through the plan is not used to train its models by default. The plan is open to labs at U.S. universities. PIs at universities without an OpenAI Enterprise agreement can apply through the trusted access process, while labs at universities with an agreement must apply through the university's Enterprise process.

   • Bringing GPT-Rosalind to academic research labs

2. Anthropic releases Code Review for Claude Team and Enterprise

   Anthropic · Products

   Anthropic has moved Code Review, a Claude Code feature that checks pull requests for bugs, out of research preview. It is available to Claude Team and Enterprise organizations. Anthropic says most reviews now finish about 3x faster than in preview, down from about 20 minutes to 6-10 minutes. Admins can set Quick, Standard or Deep review levels for each repository. Eligible Enterprise organizations can claim at least $10,000 in Code Review credits by January 15, 2027, and the credits expire 30 days after claiming. Terms apply.

   • Code Review is now generally available

3. Circle to Search on Android can now check text messages for scams

   Google · Products

   Google says Circle to Search on Android can review suspicious text messages to help users spot scams before they click a link or share personal information. Users long-press the home button or navigation bar, then draw a circle around the message. An AI Overview appears with guidance and suggested next steps. If the results show red flags, Google suggests deleting the message and blocking the number.

   • Check if a text message is a scam using Circle to Search.

4. Google expands Android earthquake alerts to more than 200 regions

   Google · Products

   Google is expanding its Android Earthquake Alerts System from 98 countries and regions to more than 200, including Panama and Guatemala. The company says it has sent over one billion alerts since 2021. Its "Be Ready" alert will gain a map of detected shaking, estimated magnitude, suggested actions and a learn-more button in the coming months. Google also published a Crisis Resilience white paper. It says its flood forecasts reach more than two billion people in 150 countries, WeatherNext can predict cyclones up to 15 days out, and fire boundary maps are live in 34 countries.

   • Android earthquake alerts now available in more than 200 regions
   • How we're helping the world prepare for natural disasters

5. Google adds more links and source context to AI Mode and AI Overviews

   Google · Products

   Google is updating its AI search features with more inline links, shown higher in responses more often. It will test a section at the top of a response that points to official pages, with context on why each link is useful. Carousels of timely articles, which highlight a user's Preferred Sources, are rolling out to more locations and languages. Google says use of Preferred Sources has grown more than 7x since the start of the year, and it started testing clickable creator names on video results in AI Mode this week.

   • More ways to connect with the web from Search

6. Google launches Universal Cart in the U.S. for Search and Gemini

   Google · Products

   Google has launched Universal Cart in the U.S. across Google Search and the Gemini app. The cart saves products from multiple retailers in one place, including items added in AI Mode in Search and in Gemini. Google says it uses the Shopping Graph's 60+ billion product listings to send alerts about price drops and items back in stock. It also uses Google Wallet to compare card perks and shows promo codes in the cart.

   • An intelligent new shopping cart

7. Meta and EssilorLuxottica start AI glasses art tour pilot at Milan's Pinacoteca di Brera

   Meta · Products

   Meta, EssilorLuxottica and Milan's Pinacoteca di Brera have launched a pilot called "Sensing Art" that uses AI glasses to make the museum more accessible to blind and low-vision visitors. The glasses connect to the museum's Grande Brera app and its DescriVedendo audio tour, which turns an artwork's shapes, composition and details into spoken words. At one of eight selected artworks, a visitor can start artwork recognition from the glasses and hear the description through the open-ear speakers. Visitors can also ask Meta AI questions or reach sighted volunteers through Be My Eyes. The tour is available by reservation, with glasses offered Tuesday through Friday from 10:00 AM to 4:30 PM.

   • Meta, Milan’s Pinacoteca di Brera, and EssilorLuxottica Expand Access to Art Through AI Glasses

8. Amazon launches Bee Frontier wearable and Bee Eclipse phone attachment

   Amazon · Products

   Amazon introduced two new Bee personal AI devices that turn conversations into summaries, to-dos and reminders. Bee Frontier is a wristband or clip-on wearable with IP67 water resistance and up to seven days of battery life, and it is available now for $74.99. Bee Eclipse is a credit-card-sized device that attaches magnetically to a phone, with more than 21 days of battery life, and it will be available soon from $99.99. Amazon says audio is processed in real time and deleted, and Bee can now connect to Alexa+ and Amazon.com. Those two integrations, along with third-party app integrations, need Bee Pro at $11.99 per month, and new customers get 30 days free.

   • Bee Frontier and Bee Eclipse: Introducing new AI wearable and phone attachment

9. GitHub makes agentic autofix for code scanning alerts generally available

   GitHub · Products

   GitHub says agentic autofix is now generally available for GitHub Code Security and GitHub Advanced Security customers on GitHub Enterprise Cloud, including the data residency option. When a code scanning alert is assigned to Copilot, it explores relevant files, proposes a fix, reruns CodeQL to check that the alert closes, and opens a draft pull request for review. Users can assign up to 25 alerts at once for a single pull request, or assign alerts through the REST API. It works only on alerts on the default branch. An enterprise administrator must turn on the new "Agentic autofix" policy, and a Copilot billing method must be set. The feature uses AI Credits from the Code Security budget and GitHub Actions minutes.

   • Agentic autofix for code scanning alerts general availability

10. Cognition launches Devin Oncall to investigate production alerts

    Cognition · Products

    Cognition's Devin has launched Devin Oncall, which investigates alerts and bug reports and helps teams diagnose and resolve production incidents. It is generally available. A responder watches a stream of alerts or bug reports, diagnoses issues automatically, and remembers underlying problems so it does not repeat investigations. Devin joins the incident channel, investigates alongside engineers, and keeps a live incident document. The company says Devin Oncall scored 100% end-to-end success on the SREGym benchmark, with the fastest time to mitigation.

    • Introducing Devin Oncall

1. Google launches Deep Review, an AI tool that checks technical documents

   Google · Products

   Google Research has released Deep Review publicly. It is an agentic AI system that checks logic and math, grounds facts and compares documents with existing work. It grew out of the Paper Assistant Tool, which gave private feedback to authors of more than 16,000 manuscripts at four computer science conferences. Google says it now also reviews design documents, presentations, and legal and financial documents. Each user gets two free reviews per week, and heavier users can connect a Gemini API key for higher limits. Google says uploaded documents are never used to train its models and are stored for one year to show in the Deep Review interface.

   • Analyzing technical documents with Deep Review

More stories

1. Google Play adds personal AI-built apps, Ask Play comparisons and Game Finder

   Google · Products

   Google says people can now use Google AI Studio to vibe code a custom Android app and launch it on their own device through Google Play, with only a standard Google account and no developer account. The rollout starts today in select markets, including the U.S., for users 18 and over. Google says sharing with friends and family is coming in the coming months, and apps shared publicly must go through a professional developer account and the same Trust & Safety reviews. Ask Play can now generate comparisons between apps, Play in Gemini will roll out to all Gemini users soon, and Game Finder, which uses a profile Google calls Gamer DNA, launches in English in the U.S. today. Google also says a redesigned Play app experience is coming to cars with Google built-in over the coming weeks.

   • New on Google Play: More ways to create, discover, and play
   • A new app experience is coming to cars with Google built-in

2. Meta adds "Your algorithm" controls to Facebook Feed and Reels

   Meta · Products

   Meta is rolling out Your algorithm, a tool that lets Facebook users add and manage topics they want to see in Feed and Reels. Users can describe their interests in their own words, and Meta says its ranking systems adapt in real time, within the same session. The tool is in the three-dot menu on every post, and Meta may ask follow-up questions after a user clicks "Interested" or "Not interested." Meta says large language models let the system understand the intent behind natural language.

   • Your Algorithm, Your Words: New Ways to Shape What You See on Facebook

3. GitHub replaces LLM calls with small classifier in secret scanning

   GitHub · Products

   GitHub says it built a two-stage secret scanner. Regex filters find candidate strings, then a fine-tuned ModernBERT classifier reads the file path, secret type and surrounding code to separate real credentials from lookalikes such as placeholders and checksums. Against the previous LLM-based pipeline, GitHub reports 11.5% higher repository owner-weighted acceptance in a customer A/B test, 86% lower median post-push scan time and more than 98% lower serving cost. On the same public traffic, GitHub says post-push scanning had 13% higher precision and 19% higher recall. The company says the faster model made it practical to extend generic secret detection to push protection, which scans changes before a push is accepted.

   • GitHub’s model for secret detection: Finding more secrets with less noise

1. Google launches Fitbit Edge and expands Google Health workout detection and Flo partnership

   Google · Products

   Google announced the Fitbit Edge tracker, which costs $179.99, is open for pre-order, and goes on shelves October 27. Google says it automatically recognizes up to 100 workouts, up from 10, and that contextual reasoning gives nearly 2x more accurate detections and cuts missed workouts by more than half. Health Guardian features, including blood pressure and insulin resistance trends, are rolling out later this month to Fitbit Air and Fitbit Edge users with a Google Health Premium subscription. Google also announced a multi-year partnership with Flo Health. Beginning early next year, users will be able to sync Flo cycle data with Google Health with their consent, and Google says that data will never be used for advertising or sold.

   • Google Fitbit Edge: Our all-new smartest fitness tracker ever
   • How we’re making Google Health and Fitbit Air smarter than ever
   • We’re partnering with Flo Health to elevate women’s health
   • Google Health: Evolving so you can focus on living better

1. ElevenLabs adds synthetic voice detection to its agent platform

   ElevenLabs · Products

   ElevenLabs has released synthetic voice detection in ElevenAgents. It analyzes a caller's speech in the first few seconds of a call and returns a determination of human or AI generated. Businesses then set their own rules, such as sending verified humans to their most capable agent, giving AI callers bounded exchanges, or stopping a synthetic voice that asks for a sensitive action like a password reset. ElevenLabs says detection does not depend on a watermark, because audio from other sources often lacks one. The company also joined the Personal Agent Protocol working group as a design partner. Meta and Sierra lead the group, which is developing rules for how a personal agent identifies itself to a business, who it represents and what it may do on that person's behalf. ElevenLabs says detection is available now to enterprise customers supported by its Forward Deployed Engineering team, and will reach a broader group of enterprise customers later this month as a configurable option.

   • Defining how AI agents interact with your business

More stories

1. Google Cloud launches Gemini agent for enterprise work

   Google · Products

   Google Cloud announced the Gemini agent at its Gemini at Work 2026 event. Google calls it a universal agent for work, used from a single prompt box for tasks such as knowledge work, answering questions, content creation and coding. Google says the agent plans the work, uses skills and tools, connects to customers' business systems, and returns finished results inside the documents, inbox and developer environments they already use. Google also says it picks the best model for each job and has built-in cost controls, plus security, administration and governance features for enterprise customers.

   • Google Cloud introduces the Gemini agent.

2. AWS releases open-source Physical AI Toolchain for building robots and smart machines

   Amazon · Products

   Amazon launched the Physical AI Toolchain on AWS, an open-source stack with architecture guidance, deployment automation and ready-to-use code. It works with NVIDIA's physical AI software and covers synthetic data, model training, simulation, edge deployment and continuous improvement. Amazon says it draws on its own robotics operations, which run more than 1 million robots, and that it helps manufacturers launch physical AI capabilities in weeks rather than years. Customers can use the full stack or pick individual components.

   • How AWS is helping companies build physical AI machines that think

3. ElevenReader launches in Brazil with Fábio Porchat voice and 70,000 Portuguese books

   ElevenLabs · Products

   ElevenLabs has launched its ElevenReader audio app in Brazil. Users can listen to books, PDFs and articles in the voice of Brazilian actor and comedian Fábio Porchat. ElevenLabs says it partnered with Bookwire Brasil to add 70,000 licensed Brazilian Portuguese titles, most of which have no audio edition. The app is free on iOS and Android in Brazil, and ElevenLabs says the full licensed catalog requires ElevenReader Ultra.

   • ElevenReader launches in Brazil with Fábio Porchat and 70,000 Brazilian Portuguese books in audio

4. Midjourney tests Thinking Mode and adds shared folders on Alpha site

   Midjourney · Products

   Midjourney is testing a "Thinking Mode" for image generation on its Alpha website. Users can click "Rerun (Thinking)" under a job's lightbox, and it works on jobs made with 8.2 (standard and edit). Midjourney says the model looks at the image, finds where it missed the prompt, and generates again. In its tests, the company says, the mode improves prompt accuracy, typography and coherence. Midjourney also launched shared folders, where owners invite people by username as Collaborators, who can generate into the folder, or Viewers, who can browse it. Folders can also be shared by link. Midjourney says sharing does not change image privacy, so non-stealth images can still appear on Explore and on profiles.

   • Testing Thinking Mode
   • Alpha Changelog - 10/7/26

More stories

1. OpenAI adds College Planner and flashcards to ChatGPT for Teens

   OpenAI · Products

   OpenAI says teens with access to ChatGPT for Teens sent about 2.7 million more learning-related messages on average than teens who did not have it yet. The company says teens spend under 15 minutes a day on ChatGPT on average, and less than 2% spend more than three consecutive hours. OpenAI will soon add College Planner, which pulls application requirements, deadlines, tasks and financial-aid steps into one plan for US students in grades 10 to 12. It is also adding flashcards, easier quiz creation, and multi-photo note capture on iOS, with Android work underway.

   • Helping teens learn, plan, and shape the future of AI

2. Google launches Playground, a platform to create and play games from text prompts

   Google · Products

   Google has introduced Playground, an experimental gaming platform where users build, play and share custom games by typing prompts, with no coding required. Games run in a browser on phones or laptops. Users can keep a game private, share a link, or publish it to the Playground Explore gallery. Google says every published game goes through safety screenings, and games support leaderboards and multiplayer in select genres. Playground launches today for U.S. users aged 18 and older, with tiered creation access rolling out based on their Google AI subscription. Google says an integration with Unity Spark, which is in testing with a closed beta coming soon, will add professional-level mechanics and 3D features.

   • Introducing Playground: Create and play custom games

3. Microsoft makes agent containers generally available on Windows 11 and adds hybrid AI features

   Microsoft · Products

   Microsoft says Microsoft Execution Containers (MXC) are now generally available on Windows 11. Organizations can set which files and networks an agent may reach, and the policy is enforced at runtime outside the agent's control. Microsoft lists support from OpenAI Codex, GitHub Copilot, OpenClaw, Replit, LM Studio, NVIDIA OpenShell and Unsloth AI, with Anthropic Claude Code, Perplexity and others to follow. Microsoft also says Windows ML now supports llama.cpp, and that HydraFusion, which lets tools use models running on the device, comes to the GitHub Copilot app, Copilot CLI and Visual Studio Code in experimental preview later in October. Microsoft says Copilot on Copilot+ PCs will get local context, local actions and local models, with rollout expected in the coming months. It also introduced a new Windows Search with thousands of actions in the taskbar, such as switching to dark mode or turning on Do Not Disturb.

   • Building Windows for hybrid intelligence
   • Microsoft Execution Containers: Policy-driven containment for AI agents

4. GitHub Copilot local sandboxing is now generally available

   GitHub · Products

   GitHub says local sandboxing for Copilot is generally available in Copilot CLI, the GitHub Copilot app, and VS Code sessions using Agent Host. Tools and commands started by Copilot run with restricted access to the filesystem, network, credentials, and other system capabilities, based on policies set by the developer or their organization. The feature uses Microsoft eXecution Container (MXC), which turns a common sandbox policy into native operating-system controls on Windows, macOS, and Linux. Organizations can use enterprise-managed settings to require sandboxing and enforce policies developers cannot weaken. GitHub says it is included with Copilot at no additional cost.

   • Local sandboxing for GitHub Copilot now generally available

5. GitHub adds ModernBERT model to push protection to block more leaked secrets

   GitHub · Products

   GitHub says it built a ModernBERT classifier that judges candidate secrets in context and evaluates batches in under two milliseconds. GitHub says adding the model to push protection more than doubles the number of secrets it can prevent. Today push protection stops about 30% of newly detected secrets before they enter repository history. The feature is in private preview, and GitHub says it will reach organizations with GitHub Secret Protection on Enterprise Cloud and GitHub Teams later this month, and that it will consume AI credits. It will also ship in public preview with GitHub Enterprise Server 3.23, including air-gapped environments.

   • Secret protection must scale with software

1. Claude adds add-on and connectors for Google Docs, Sheets and Slides

   Anthropic · Products

   Anthropic has released Claude for Google Workspace in public beta on all paid Claude plans. The add-on opens Claude in a sidebar next to an open file, where it can read the document, see the selected text, cells or slides, and edit the file. In Sheets, Anthropic says Claude can write formulas, build pivot tables and charts, and add tabs. In Slides, it can build slides from the deck's layouts and themes. By default Claude asks for approval before each edit, and an "Accept all edits" mode applies changes without stopping. Anthropic also released Google Docs, Sheets and Slides connectors, also in beta, so users can create and edit Google files from a Claude chat. On Team and Enterprise plans, an owner must enable the connectors first.

   • Claude now works with Google Docs, Sheets, and Slides

2. Google Flow Music lets producers export custom tools as VST3 and AU plugins

   Google · Products

   Google says producers can now export custom Spaces from Google Flow Music as VST3/AU plugins that run inside their digital audio workstations. Spaces lets creators build custom instruments or effects using natural language, with no coding experience required. Google gave one example: producer Khris Riddick-Tynes built a plugin called "No Chaser" to keep his instrumentals and vocals sharp and studio-quality, no matter where they were recorded.

   • Producers can now vibe code their own music production tools using Google Flow Music.

3. ElevenLabs launches ElevenAgents Architect to help teams improve voice agents

   ElevenLabs · Products

   ElevenLabs has released ElevenAgents Architect, a built-in expert that helps teams launch and improve agents by voice or chat. ElevenLabs says it analyzes transcripts across thousands of conversations, proposes changes, and writes simulations to validate them. Every change lands as a versioned draft that can be reverted, and approval is required before anything goes live. The company says teams can also reach it from Claude, Claude Code, ChatGPT, Cursor, and Grok Bot. It is available now in Alpha.

   • Introducing ElevenAgents Architect

4. Cohere launches North 2 agent platform and PwC alliance

   Cohere · Products

   Cohere released North 2, which it calls its biggest upgrade yet to its enterprise agentic AI platform. It adds a redesigned agent orchestration system, plus skills, libraries, memory and natural-language app creation. The platform works with Cohere's models or a customer's own, and it can run in the cloud, on-premises or fully air-gapped. A North Admin tool sets cost controls, rate limits, user quotas and guardrails at the agent level. Cohere also announced a global alliance with PwC, launching first in Canada, to help organizations deploy North and Cohere's models and search tools.

   • North 2: Enterprise AI without compromises
   • Cohere and PwC partner to accelerate secure enterprise AI adoption

1. Devin adds Memory and Dreaming to retain lessons across sessions

   Cognition · Products

   Cognition has added Memory and Dreaming to Devin. Memory lets Devin keep your preferences, corrections and project lessons from one session to the next. Dreaming is a daily background process that merges overlapping notes, removes stale ones and looks for lessons missed during the original work. Memories are stored as markdown notes in a personal Git-based Memory Drive, are personal to each user within an organization, and can be browsed under Customize → Memory. Cognition says it also open sourced the standard it used to build the memory system.

   • Memory and dreaming: how Devin learns from working with you

1. OpenAI releases GPT-6 Astra Ultrafast on NVIDIA Blackwell GPUs

   NVIDIA · Products

   GPT-6 Astra Ultrafast is available now in the OpenAI API and to eligible ChatGPT Work and Codex users, according to NVIDIA. It runs on NVIDIA Blackwell GPUs and offers up to 8x faster token generation than the Astra Standard mode. NVIDIA says faster generation can shorten the edit-test-debug cycles of coding agents. OpenAI's Philippe Tillet says the company's models are good at writing high-performance kernels for Blackwell and Rubin GPUs.

   • How NVIDIA GPUs Help Accelerate OpenAI’s GPT-6 Astra Ultrafast