Scoop
AI tool news · rumor vs. reality

The AI Wire ●

Rumors tracked. Announcements verified. Updated daily.

Showing Released · show everything

2026-10-28

RELEASED

Grok Bot 0.77 rolls out with a Permissions page, a Payments tab in the Vault and multi-question cards

Grok Bot 0.77.0 is stable and rolling out, according to its changelog coverage on October 28. Settings gains a Permissions section that lists a user's Auto-review rules under Always Ask and Allow without Asking.

The Vault gets a Payments tab where users pick which payment method Bots try first, though every purchase still needs approval. Bots can now ask several questions on one card that steps through them, with room for the user's own answer.

The changelog also lists 1Password logins without a shared vault, Import Memory from another AI assistant, and approval cards that stay open 30 minutes instead of 10. New Order Concierge and Deal Watcher templates round out the release.

RELEASED

Shopify's Fall '26 Edition turns AI agents into a sales channel: Shop.md, and Shop Pay inside Muse, Claude Duet and Grok Bot

Shopify released its Fall '26 Edition, 'Exponential,' on October 28 — more than 150 updates, with agentic commerce at the center. The headline for shoppers is Shop.md, a shopping skill buyers can point any AI agent to: it searches more than a billion products, uses Shop preferences and order history for recommendations, checks out with Shop Pay and tracks orders.

Shop Pay checkout now also runs inside personal AI agents, starting with Meta's Muse, Claude Duet and Instinct. Claude Duet, Grok Bot and Instinct go a step further and can offer Shop Pay for brands that are not on Shopify at all, using card-network tokens from Mastercard, Visa and Discover — American Express is listed as coming soon — so the buyer's card is never shared with the agent.

For merchants, Shopify's connectors now work with nine AI agents — ChatGPT, Claude, Claude Duet, Microsoft Copilot, Google Gemini, Grok, Grok Bot, Muse and Perplexity — letting them manage products, inventory, orders and performance from a conversation. Shopify says more than 400 million people already use Shop Pay.

RELEASED

OpenAI now lets you text your Dot agent over iMessage and RCS

OpenAI has begun rolling out texting for its Dot agents, nearly a month after Dots debuted with a promise that users would eventually be able to text them. The preview is rolling out among US Pro users first.

You can send your Dot texts, photos and even voice memos, and it can text you back proactively with updates on tasks you have assigned to it. Setup lives in the Dot's profile in ChatGPT under Texting, and OpenAI says memory is shared across ChatGPT and text, so context carries over between conversations.

OpenAI says it partnered with Linq to provide messaging support and did extensive penetration testing, stressing that you are the only one who can text your Dot, and your Dot can only text you. The launch sharpens the always-on agent race with Anthropic's Claude Duet, which is still waitlist-only in the US.

RELEASED

OpenAI offers up to $500 a month in API credits to eligible ChatGPT Pro and Business subscribers

OpenAI announced on October 28 that eligible ChatGPT Pro and Business subscribers can claim up to $500 per month in API credits to build apps and agents. The offer came from the company's @OpenAIDevs account, with the eligibility rules set out in an updated Help Center page.

To qualify, a subscriber needs a Pro subscription, or a Business workspace with paid seats, that has been active for at least seven days, and must own the API organization that receives the credits. Availability also depends on the country.

The credits are a subscription perk, not extra ChatGPT capacity. They fund a single linked API organization and do not extend ChatGPT usage or Codex usage on account sign-in, and API billing stays separate from the ChatGPT bill.

RELEASED

Nvidia releases Nemotron 3.5 Transcribe, a free open-weight model for audio in up to 40 languages, plus Granary dataset v2

Nvidia on Wednesday, October 28 released Nemotron 3.5 Transcribe, a free, open-weight model that transcribes audio in up to 40 languages. It also released version 2 of Granary, its open source speech dataset.

Nvidia says Nemotron 3.5 can translate 31 languages to and from English, handle recordings up to an hour long, and tell up to eight speakers apart. It claims the model beats rivals on benchmarks including Hugging Face's Open ASR Leaderboard, and targets developers building apps for business meetings and medical transcription.

One drawback is that Transcribe works with recorded audio, not livestreams. At 30 billion parameters, Nvidia says it can run on a single chip, from the Jetson AGX Thor robotics controller and DGX Spark to Blackwell datacenter chips.

RELEASED

Meta launches Muse in Singapore, its fourth market in 50 days, with Shopee, Grab and Singapore Airlines among local partners

Meta made Muse, its personal AI agent, available in Singapore on October 28, on iOS, Android and the web at muse.ai. Singapore is Muse's fourth market in 50 days since its September debut in the US and Canada, and its second market outside North America after Japan on October 23.

From the start, Muse links to Gmail, Google Calendar and Instagram, so it can sort email, arrange schedules and draw on saved posts. Meta named eight local partners whose apps Muse will connect to, among them Shopee, Grab, Lazada and Singapore Airlines, plus Agoda, ShopBack, Klook and Trip.com.

Meta's examples include comparing prices before buying on Shopee or Lazada, planning groceries and meals through Grab, and earning cashback through ShopBack. The company says partner features will roll out progressively, with some arriving after launch, without saying which work on day one.

RELEASED

Google adds Immersive Explorations to Gemini, letting students explore science and history in interactive scenes

Google has introduced Immersive Explorations in the Gemini app, a feature that lets students explore topics from microscopic systems to deep space through interactive scenes. The experiences are aimed at science and history topics that are hard to picture from text alone.

Students can open the library from the Gemini student hub by choosing "Try immersive explorations," and Google says the experiences will also be accessible directly inside Gemini chat conversations soon. Audio narration to guide users through the scenes is also on the way.

Immersive Explorations is available in English, Indonesian, Hindi, Brazilian Portuguese, French and Simplified Chinese. Google announced the feature in a blog post on Wednesday, October 28.

RELEASED

Chrome's AI autofill can now pull flight bookings and order numbers from your Gmail

Google said on Wednesday, October 28 that Chrome's enhanced autofill can now suggest details pulled directly from a user's Gmail while they fill out online forms. The suggestions can include order numbers, package tracking information, flight bookings and hotel reservations, surfaced under a 'Suggested by Gemini' section.

The feature is powered by Gemini with Personal Intelligence and is opt-in: it is rolling out in the US to Gemini in Chrome users who have Personal Intelligence enabled, on desktop, Android and iOS. Google says sensitive data stays encrypted, and users can remove suggestions or turn the feature off in Chrome settings.

It builds on last year's autofill upgrade, which could fill in details such as passport numbers and addresses from a Google Account and Google Wallet. Google says Google Photos is next, with autofill soon able to pull information from images stored in a user's Photos library.

RELEASED

GitHub makes Copilot Fusion generally available: it picks and combines models so developers don't have to

GitHub used its Universe 2026 conference on October 28 to make Copilot Fusion generally available. Instead of picking a model themselves, developers select Fusion in the model picker and choose an effort level — Standard for everyday work, High for harder tasks, or Max for the most demanding work — and Copilot decides which models to use and whether several should work together.

Fusion combines Copilot's automatic model selection with the multi-model orchestration from Project HydraFusion, which arrived as a research preview in September. Depending on the task it can run a single model, cascade from a cheaper model to a stronger one after a quality check, or have one model critique another's output.

GitHub says Fusion Max scored within two points of Claude Code running Opus 5.5 Max on the Artificial Analysis Coding Agent Index, at about half the cost and 40% less time per task. Enterprise administrators can restrict which models Fusion may draw on and can set it as the default for their users.

RELEASED

Anthropic releases a ready-to-use artifacts gallery for Claude for Legal, built with law firms and legal-tech partners

Anthropic on Wednesday, October 28 introduced a gallery of ready-to-use artifacts for legal work in Claude, extending the Claude for Legal product it launched in May. The gallery brings together six Anthropic-built task artifacts, four artifacts built by legal tech companies and law firms, and nine access-to-justice artifacts.

Artifacts are interactive tools such as trackers and dashboards that teams can use and share inside Claude. Every artifact in the legal gallery opens with an example so users can see the workflow before connecting their own data, and every output links back to its source.

Anthropic said the most common requests from legal teams since the May launch were for simpler setup and easier ways to work with colleagues. Under the gallery's working model, the lawyer supplies the documents and sources, Claude produces a first pass, and the lawyer then applies judgment to the output.

RELEASED

Anthropic ships Claude Enterprise cost tools: smart reports, in-app surveys and pooled group budgets

Anthropic made three Claude Enterprise features generally available on October 28: smart reports, in-app surveys and pooled group budgets. The launch is aimed at administrators who need to show how employees use Claude, what it costs, and whether it is proving useful.

Smart reports, in beta since September 10, let an admin pick a team, the Claude products to include and a period of up to 28 days, then analyse samples of that team's transcripts. The reports group the work into workstreams, show what was produced and the cost at standard pricing before discounts, and flag problems such as failed tool calls and workflows that could become reusable skills; each organisation gets 10 reports a month free, and Anthropic says they are not for evaluating individual employees.

Pooled group budgets give a team one shared monthly spending limit on top of each member's own limit, with use pausing when either limit is hit and billing admins alerted at 50%, 75%, 95% and 100%. In-app surveys ask employees what Claude helped them achieve, and admins can edit the questions and export the answers.

RELEASED

Anthropic quadruples the Claude API request-size limit, from 32 MB to 128 MB

Anthropic raised the maximum request size for the first-party Claude API's Messages API and Token Counting API from 32 MB to 128 MB on October 28, according to its developer-platform release notes. The change is live for developers on the first-party API.

The cap measures the raw bytes of the HTTP payload, not tokens, and it bit hardest on image-heavy requests: encoding images and PDFs for transport inflates their size by roughly a third, so developers could exhaust the old 32 MB ceiling — and hit hard 413 errors that stopped entire conversations — while still having plenty of context window left. Longer computer-use sessions are another beneficiary.

The increase does not change pricing or the context window itself. The other API surfaces keep their separate limits: 256 MB for the Message Batches API and 500 MB for the Files API.

2026-10-27

RELEASED

OpenAI adds iCloud Mail and Calendar plugins to ChatGPT, on all plans across web, desktop and iOS

OpenAI launched iCloud Mail and iCloud Calendar plugins for ChatGPT on Tuesday, October 27, letting users search their inbox, catch up on messages, check what is next and update calendar events in natural language. The plugins are available on all ChatGPT plans, free and paid, across web, desktop and iOS.

Connecting uses an app-specific password generated from the user's Apple Account, which requires two-factor authentication, rather than the account's main password. The plugins talk directly to iCloud's services rather than to Apple's desktop Mail and Calendar apps, so third-party accounts managed only inside those apps stay out of reach. On the Pro tier, OpenAI's autonomous agent Dots can also use the plugins to work with iCloud data.

Tibo Sottiaux teased the launch in his Day 23 challenge post — “Much requested iCloud Mail & Calendar plugins. Just kidding, today is a reset. Still much requested though!” — before the official ChatGPT account confirmed the plugins were live.

RELEASED

Microsoft Copilot Memory now learns from your emails, Teams chats and meeting transcripts — 'work insights' are rolling out, and you can turn them off

Microsoft is rolling out a Copilot personalization update for eligible Microsoft 365 business and education accounts, Forbes reported on October 27. Copilot Memory now uses 'work insights' — details it infers from your Outlook emails, Teams chat records and meeting transcripts — to personalize its responses around your preferences, relationships and work priorities.

Microsoft says the data stays confidential and is not used to train its models. Users stay in control in three ways: ask Copilot 'What do you know about me?' for a summary of what it uses, ask it to forget a specific detail, or turn the whole thing off under Settings > Chat Settings > Personalization by disabling 'Chat History and Work Insights'. Once disabled, the inferred memories are deleted after 30 days — and may be restored if the setting is switched back on inside that window.

The feature is Microsoft 365 Roadmap item 551195, scheduled for early to late October 2026, and the Forbes report indicates the rollout is now reaching tenants. It lands in the same month as fresh outside audits of exactly this kind of inference, which found Copilot's self-reported memory list may be incomplete and that individual wrong or unwanted insights cannot always be removed one by one — making the all-or-nothing toggle the practical control.

2026-10-26

RELEASED

OpenAI releases GPT-6.1 Terra at $1/$5 per million tokens, half the price of Sol, for ChatGPT Work, Codex and the API

OpenAI released GPT-6.1 Terra on Monday, October 26, the mid-tier model that was left out when the GPT-6 family launched last month. It succeeds GPT-5.6 Terra and is available now in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and through the API as gpt-6.1-terra; it is also generally available on Amazon Bedrock. OpenAI positions it for workday computer use: drafting proposals, reading technical drawings, filling out forms and editing video.

The price is the headline. API pricing is $1 per million input tokens, $0.05 per million cached input and $5 per million output, which OpenAI says is half the price of both GPT-5.6 Terra and GPT-6.1 Sol. The context window is 1.05 million tokens with up to 128,000 tokens of output, and prompts above 272,000 input tokens are charged at higher rates.

On OpenAI's own figures, Terra scores 50.3% on Terminal-Bench 4.0 at maximum effort against 21.5% for GPT-5.6 Terra, and its error rate on an internal difficult-prompts test fell from 53.6% to 16.7% at low effort and to 4.3% at maximum effort. OpenAI also says it completes tasks more than three times faster than GPT-6 Astra and GPT-6.1 Sol, while still recommending Sol and Astra for the most complex coding. Artificial Analysis calls it a new mark for intelligence per dollar; those benchmark figures are vendor-reported and early third-party results, not yet independently settled.