Scoop
AI tool news · rumor vs. reality

The AI Wire ●

Rumors tracked. Announcements verified. Updated daily.

2026-11-13

RELEASED

Google rolls out Dynamic Workflows in Antigravity: /run-workflow turns a task into a Python script that runs the agents

Google is rolling out Dynamic Workflows in Antigravity, its agent coding platform. Type /run-workflow and a task, and the main agent writes a short Python script that coordinates other agents — parallel() runs one agent per file or package at once, pipeline() runs them one after another — with each step returning a structured result checked against a JSON schema.

The point is deterministic control flow instead of prompt drift: the script is reviewable before it runs and can be saved and re-run, and each workflow-launched agent works in its own conversation, without seeing the orchestration script or its peers. The generated programs use a small asynchronous API covering phases, logging, agent creation and parallel or sequential execution.

Dynamic Workflows are available with Google AI Pro and Google AI Ultra in the Antigravity app and CLI, with no separate price — runs consume the plan's existing model credits — and workflows execute in a separate process using Antigravity's bundled Python runtime. Google announced the rollout on November 13.

Update, November 14: Antigravity lead Varun Mohan said the team is resetting everyone's Gemini quota on Antigravity to mark a week of steady shipping that also included Hydrogen, Orchestrator and Knowledge work. Neither post announcing the reset says how long it lasts, so developers should treat it as a short window. The same day, Antigravity CLI 1.5.9 shipped: parallel subagent workflows can now run with no Python install, organisations get new plugin controls, Auto mode is faster, and a fix stops MCP headers leaking on redirects.

ANNOUNCED

GitHub Copilot will retire six models on December 14 — Fable 5, Haiku 4.5, Gemini 3.8 Flash, Gemini 4 Argon, Grok 4.6 and Kimi K3

GitHub announced on November 13 that it will deprecate six models across all Copilot experiences — Chat, inline edits, ask and agent modes, and code completions — on December 14, 2026. The models are Claude Fable 5, Claude Haiku 4.5, Gemini 3.8 Flash, Gemini 4 Argon, Grok 4.6 and Kimi K3.

GitHub's own changelog table names a suggested replacement for five of them: Fable 5.2 for Fable 5, Haiku 5.5 for Haiku 4.5, Gemini 4 Hydrogen for Gemini 3.8 Flash, Gemini 4 Carbon for Gemini 4 Argon, and Grok 4.7 for Grok 4.6. Kimi K3's replacement is listed as “to be announced.” Some secondary roundups misprint the Fable replacement as Fable 5.1; GitHub's table says 5.2.

One exception applies: Claude Haiku 4.5 stays available to individual Copilot subscribers on annual plans. For Business and Enterprise customers the replacements are enabled automatically under default model enablement, and GitHub says nothing needs to be done to remove the old models once the date passes.

The same day, GitHub made Gemini 4 Hydrogen generally available in Copilot — the third new model in its picker in four days, after Claude Fable 5.2 on November 9 and Gemini 4 Carbon on November 10. Teams that have pinned a workflow or config to one of the six retiring model names need to switch by December 14 or lose the model mid-workflow.

ANNOUNCED

Anthropic publishes ten recommendations for building safety evaluations — and says it has paused many tests that give models internet access

Anthropic published a research post, “Building strong safety evaluations,” on November 13, setting out ten recommendations for designing, running and reporting the evaluations that decide whether a model is ready to deploy, which safeguards ship with it, and what the next round of safety work should target. Anthropic's premise is blunt: researchers keep finding that AI evaluations are easy to game, do not always measure what they claim, and are not documented well enough to reproduce.

The practical recommendations include preregistering how results will be analysed before seeing the data, validating automated graders against human expert judgement, and moving beyond single-turn refusal rates to multi-turn tests that report whether the model behaved appropriately throughout a conversation. Anthropic also argues for multidisciplinary evaluation teams — methods experts alongside domain specialists such as psychologists and biologists — and for keeping a severe flaw private until a fix is possible.

The post also discloses a change to Anthropic's own practice. After incidents in its own evaluation work in recent months, the company says it has hardened its evaluation environments, temporarily stopped running many evaluations that give models internet access, and put the web-search evaluations that still need a connection under real-time monitoring, with runs sandboxed and stopped immediately if a model acts in ways that could cause real-world harm.

RELEASED

Anthropic upgrades Claude Code Projects: the coordinator now answers in project chat, threads can be shared read-only

Anthropic announced a round of updates to Claude Code Projects late on Friday, November 13, in a @ClaudeDevs thread. The coordinator — the chat that scopes work and delegates it to parallel worker threads — is now, in Anthropic's words, “much more proactive”: it answers questions directly in the project chat and posts status updates on running threads instead of waiting to be checked on.

Individual threads can now be shared as read-only links with anyone signed in to Claude, with one exception: threads that had access to your computer cannot be shared. Projects also gain reminders that hide a thread until a set time, and /btw — which spins a tangent into its own side thread — and /fork, which branches a new thread from the current one, now work inside projects.

The threads panel adds a recents row and artifact thumbnails, and Anthropic caps accounts at 200 new threads per day across all projects. Projects remains in public beta on Pro and Max, with a waitlist for everyone else; the redesigned, coordinator-led version launched on September 17.

RELEASED

Claude Code on desktop gets custom themes: mods can now recolour the app and add animated backdrops

Anthropic's ClaudeDevs account has announced custom themes for Claude Code on desktop, built through the mods plugin system. A new $.theme API lets a mod set an app-level backdrop — static or animated — while ui.render hooks restyle individual interface elements and colours.

Users do not have to write the theme themselves: they can describe the look they want and let Claude build the mod, or start from Anthropic's theme-examples plugin. A community marketplace already packages familiar palettes such as Dracula, Catppuccin, Gruvbox, Nord and Tokyo Night.

The feature is live for local desktop sessions on Claude Code 2.1.319 or later, and mods are included with Claude Code at no extra charge. Cloud sessions are listed as coming soon, with no date announced.

RELEASED

Claude Code 2.1.325 patches a zstd decoding security issue and makes sandboxed Bash commands ask permission by default

Anthropic released Claude Code v2.1.325 on November 13, a day after the Code Mode release. Its headline item is a security fix, stated in a single sentence in the release notes: “Fixed a security issue in how zstd-compressed responses are decoded.” The notes name no CVE, list no affected versions and describe no exploit path, so the honest description stops there.

The second change is a default flip with practical bite: sandboxed Bash commands now follow the normal permission flow by default instead of running without a prompt. Users who want the old behaviour set sandbox.autoAllowBashIfSandboxed to true — and where the sandbox is switched on through managed settings, that setting has to go in the highest-ranking managed source.

The release also fixes subagents with their own model being moved to the plan-mode model under opusplan or haiku settings, and fixes compaction on many third-party gateways and on Microsoft Foundry, alongside a longer list of auth and MCP fixes. Two releases in two days that mostly close permission gaps are a good reason to update before the next unattended run.

2026-11-12

RELEASED

Grok Bot on Mac can now use Messages, Mail, Notes and four more Apple apps — and you can drive your Mac from your phone, the web or a Tesla

The official Grok Bot account announced on November 12 that Grok Bot on Mac can now use seven Apple apps and services: Messages, Mail, Notes, Reminders, Calendar, Location and Contacts. Each one gets its own Allow button on a new Connect Apps screen, so access is granted app by app rather than all at once.

The apps do not all get the same powers. Messages is the strongest: the bot can read conversations and send messages for you. Mail is search and drafting, with no sending shown; Notes can be read and written; Reminders, Calendar, Contacts and Location cover checking and adding the everyday records those apps hold. Grok developer Evan Bacon added the reach: you can remote-control your Mac from any Grok Bot client — phone, web or a Tesla — to check a calendar, send a message or read reminders.

The Mac list is longer than the phone list. Grok Bot added Calendar, Reminders, Contacts and Location on iPhone on October 22, but Mail, Notes and Messages are Mac-only for now — iMessage access in particular has to run through a Mac. Two honest bounds: xAI had published no help page for the Mac connections when reporters checked on November 13, and the announcement does not state a plan requirement or a minimum macOS version, so secondary reports that name one should be read as unconfirmed.

ANNOUNCED

SpaceX plans more than $74B in AI supercomputing campuses in Texas and Arizona

SpaceX said on November 12 that it plans to invest more than $74 billion in two AI supercomputing campuses: more than $24 billion at its Starbase site in Texas and roughly $50 billion in Cochise County, Arizona, Reuters reported from a company statement. SpaceX's AI segment — the business that now includes xAI's Grok — generated more than half of the company's third-quarter revenue, underlining how far the company has moved beyond rockets and Starlink.

Neither campus is approved yet. The Arizona campus needs Cochise County approval: it would run to about 6 GW at full build-out, use dry cooling with no planned industrial groundwater use, and work with local electric cooperatives in a deal SpaceX says could cut members' bills by up to 20%. A petition with 2,400 signatures asks officials to deny the permit, and a county hearing is set for December. SpaceX estimates about 5,500 construction jobs and 1,000 permanent jobs in Arizona.

The Starbase campus would generate its own electricity rather than draw from the public grid, with a city notice showing a gas-fired power plant and supercomputing facility on about 444 acres. The Starbase City Commission is due to consider a property-tax abatement on December 9. SpaceX says the Texas project has already generated more than 3,000 construction jobs and would support more than 500 permanent jobs once operating.

RELEASED

OpenAI launches local API processing in India: GPT-6 Luna inference residency for eligible businesses

OpenAI announced on November 12 that eligible businesses and developers in India can now process API requests entirely in-country, extending its existing India data-residency options from storage to inference. At launch the offer covers GPT-6 Luna only, through the Responses and Chat Completions APIs, via an India regional endpoint on a new API project set to India for storage and processing.

India becomes the third market with inference residency, joining Australia and Japan. Regional processing in each requires Zero Data Retention to be enabled and Standard pay-as-you-go processing, and hosted tools are not supported under the in-country setup. The launch sits inside the OpenAI for India initiative spanning infrastructure, enterprise adoption, skills and local partnerships.

The scope is deliberately narrow: no other models were named for India processing, and OpenAI gave no timetable for extending it and no investment figure or data-center details. For regulated Indian customers, though, keeping the full request cycle — receipt, computation and result — inside India is the practical change that unblocks workloads that storage residency alone did not.

ANNOUNCED

OpenAI launches Daybreak Red Team and Defense Factory: agents that pen-test web apps and draft the patches

OpenAI expanded its Daybreak cybersecurity platform on November 12 at its Cyber at the Frontier keynote, adding two products built on its cyber models, currently GPT-6 Cyber. Daybreak Red Team, in early access by application, sends coordinated agents that use source code and application context to explore attack paths in web apps, validate reported vulnerabilities and check whether fixes hold. President Greg Brockman framed it bluntly: companies will be secure only to the extent of the frontier model they use for defense.

Daybreak Defense Factory covers the fixing half. Its agents take findings from Red Team or from tools a customer already runs — Wiz, Snyk and Endor Labs are named — trace each vulnerability through the codebase and prepare merge-ready patches with tests, which engineers review before anything merges. It ships as a preview plugin inside ChatGPT for workspaces that already have Daybreak access.

OpenAI published no pricing and no independent performance results for either product. A benchmark chart circulated from the event showing an unreleased model at 77% has not appeared on OpenAI's product pages, so treat those numbers as unverified.

ANNOUNCED

OpenAI's proposed Cursor cutoff date arrives today — built-in GPT models were set to go dark, with talks still going on November 11

Today, November 12, is the proposed shutoff date OpenAI set on August 28 for ending its direct supply of models to Cursor, the AI code editor SpaceX acquired in a $60 billion deal completed in August. OpenAI said at the time it could not be confident SpaceX would use its technology within its terms of service, citing a pattern of contract breaches by Musk companies, and that it would not supply future models to Cursor either. It framed November 12 as a proposed date and gave the longest notice its contract allowed.

What changes if the cutoff takes effect as proposed: the built-in OpenAI models disappear from Cursor's model picker, while Claude, Gemini, Grok and Cursor's own Composer models stay. Developers can still use their own OpenAI API key in Cursor's local Chat and Agent, and OpenAI's own IDE extensions keep working inside Cursor. Cursor CEO Michael Truell has said OpenAI models account for about 5% of Cursor's user traffic.

The outcome is not confirmed. As of November 11, reporting still described the two companies as talking to try to resolve the dispute, and neither side had said whether the models were still live that day. Treat today as the deadline arriving, not as proof the models have actually gone dark — confirmation from Cursor's picker or either company is the next signal.

Update, November 12 evening: the date has passed its morning without a confirmation either way. As of 9 a.m. Central Time on November 12, Enjoy.dev found that neither company had confirmed the cutoff, Cursor's models page still listed GPT-5.6 Sol, Terra and Luna with no November 12 note, and OpenAI's help page still called November 12 a proposed transition date, saying the official termination date would be shared once both companies confirm it. What is settled either way: OpenAI's current GPT-6 family is not in Cursor and will not be supplied under these terms.

Update, November 13: the cutoff did not take effect. On November 12 OpenAI added a line to the top of its original post: 'We've temporarily extended Cursor's access to our models beyond November 12. We will provide updates as available.' No new end date, reason or duration was given. As of November 13, Cursor still lists GPT-5.6 Sol, Terra and Luna, OpenAI's help page still calls November 12 a proposed date, and Cursor has made no public statement we could find. GPT access in Cursor continues for now, on terms that could still end on short notice.

RELEASED

OpenAI brings ChatGPT Finances to the Mac desktop app: linked bank accounts, spending and portfolios in the sidebar

OpenAI announced on X that ChatGPT Finances is now available in the ChatGPT desktop app on Mac. After updating the app, US users on Free, Go, Plus and Pro plans can open Finances from the sidebar and connect bank, credit card and investment accounts through Plaid.

Finances first launched in May as a Pro preview and has since expanded to Plus, then Free and Go users in October; it was already on web, iOS and Android, and gained voice support earlier this month. The dashboard shows bills, subscriptions, net worth and investments across more than 12,000 supported institutions, and users can ask questions about transactions and compare spending across months.

The connection is read-only: ChatGPT cannot transfer money, pay bills or place trades, and accounts can be disconnected at any time. OpenAI has not announced a Windows desktop date or a rollout beyond the US.

ANNOUNCED

OpenAI expands ChatGPT Ads buying to 44 countries through an exclusive Aleph reseller deal

OpenAI is widening how ChatGPT Ads are bought. Under an exclusive reseller partnership announced Thursday, November 12, brands and agencies in 44 countries can now contract, pay and get local support for ChatGPT advertising through Aleph, rather than dealing with OpenAI directly.

The footprint covers 14 markets in Latin America, 16 in Europe and Central Asia, 10 in the Middle East and Africa, and 4 in Asia-Pacific. It follows the October expansion that took ChatGPT ads live in 89 markets and the November 9 switch to Target CPA and Target CPC bidding.

This is a sales-channel change, not a new ad format: what users see in ChatGPT does not change, but local-currency, local-language buying makes the ad business considerably easier for advertisers outside OpenAI's direct markets to join.

ANNOUNCED

Researcher finds Nvidia Blackwell fingerprints in six Chinese open models — DeepSeek, Kimi, MiniMax and Ant Group — but no violation is established

Will DePue, a former OpenAI researcher who leads the nonprofit Unemployment Labs, published a forensic report on November 12 arguing that the published weights of six Chinese open models carry the fingerprint of Nvidia Blackwell GPUs — chips that have required a US export licence for China since before they shipped. The method uses how PyTorch scatters random starting values across a chip's compute cores: Hopper chips have 132, Blackwell B200/B300 chips 148, and Grace Blackwell GB200/GB300 systems 152, and some starting values survive training untouched.

The report reads DeepSeek-V4-Flash and MiniMax-M3 as B200/B300-class, and DeepSeek-V4-Pro, Moonshot's Kimi K3 and K3.1-Pro, and Ant Group's Ling-3.1-flash as GB200/GB300-class, with reported statistical strength from 49 to 1,385 sigma. Earlier models from the same four labs read as Hopper, and DePue says the method correctly identified all 24 control models whose training hardware is public. He estimates roughly 12,000 Blackwell GPUs across five inferred runs, while saying he is not confident in those figures.

The bounds matter as much as the finding. The fingerprint identifies the hardware that generated a model's starting weights; training on the same hardware is an inference, not a direct observation. It also says nothing about where the chips were or who owned them — Chinese labs can rent compute in data centres abroad, and one cited estimate has large numbers of accelerators reaching the region through third countries. This is evidence of access to export-controlled hardware, not by itself proof of smuggling or an export-law violation, and none of the four labs has commented in the coverage so far.

Update, November 15: the case got a cooler second reading. DePue published a more detailed forensic write-up of the method, and the Hacker News response was notably more sceptical than the first coverage: commenters pointed out that renting GPU time on Blackwell machines physically located in Singapore or Malaysia is entirely legal under US export controls, so a fingerprint in the weights does not by itself imply a violation. The follow-up also notes that DeepSeek's newer V4.1-Flash showed no Blackwell signature at all. The original bounds stand — the fingerprint identifies the hardware behind a model's starting weights, not where the chips were or who owned them.

ANNOUNCED

Nscale's amended IPO filing: no binding financing yet for its $44.6B Anthropic campus — and Anthropic has exercised its second-building option

Nscale, the London-based, Nvidia-backed cloud company building Anthropic's Monarch Compute Campus in Mason County, West Virginia, said in an amended IPO filing on November 12 that it has placed large orders for GPUs and networking gear to meet the terms of its agreements with Anthropic — orders that cannot be deferred, reduced or cancelled — and that it "will be required to obtain additional financing" to meet its 2027 obligations. The August agreements, covering Nvidia Vera Rubin systems at Monarch, could pay Nscale up to about $44.6 billion, depending on delivery conditions.

The filing puts numbers on the gap. The first Monarch building is expected to cost $10 billion to $10.5 billion, plus another $21 billion to $22 billion for GPUs and computing equipment, and VKTR reports Nscale has no binding financing yet for the buildout and estimates it still needs $9 billion to $9.5 billion to finish that first building. Nscale's total debt reached about $5.5 billion on September 30, up from $137.4 million three months earlier. Its preliminary nine-month figures are estimates its auditor has not reviewed.

The same filing moves the rumoured second deal forward without closing it. Anthropic exercised its right on September 29 to contract for more compute in a second building at Monarch, which opened a 60-day window for the two companies to agree terms — the formal version of the talks Bloomberg reported on October 19. No second agreement has been signed. Anthropic's own prospectus already warns that some compute partners rely on outside financing and could deliver late, or not at all.