The latest wire

Vercel added Xiaomi’s MiMo V2.6 Pro, Flash and Pro UltraSpeed to AI Gateway. All three accept text, images, audio and video, with a one-million-token context window and up to 128,000 output tokens. Pro targets complex work, while Flash is positioned as a lower-cost option. Vercel says UltraSpeed serves the same Pro model with up to twentyfold faster output.

Vercel added Grok 4.7 to AI Gateway for extended coding, document and presentation work. The model offers a 500,000-token context window and four reasoning levels, letting applications choose how much reasoning to request. It can also be selected in Vercel’s eve agent framework. The launch promotion offered 40% off only through September 27, so that advertised discount has expired.

You can now grant teammates scoped access to specific Workers directly from the Workers dashboard. Go to your Worker and click Invite. Enter the teammate's email address, choose the appropriate access level, and click Invite. You can grant a user one of the following access levels: Metadata Read-Only: View settings, metrics, logs, and traces without access to Worker code or the ability to make changes. Content Read-Only: Read Worker code, settings, and observability data without the ability to modify or deploy changes. Editor: Update and deploy a Worker without the ability to delete it. Admin: Everything included with Editor, plus the ability to delete the Worker. If the teammate is already an account member, they will receive access to the Worker immediately.

De multiples vulnérabilités ont été découvertes dans Synology DSM. Certaines d'entre elles permettent à un attaquant de provoquer une exécution de code arbitraire à distance, un déni de service à distance et une atteinte à la confidentialité des données.

Sep 20, 23:22 UTC Resolved - On September 20, 2026, between 21:46 and 22:24 UTC the Pull Requests service was degraded and pull request merge and test-merge commits were created late, with delays reaching approximately four minutes at peak.

EFF has published a guide to limiting which personal information Apple's updated Siri can use in iOS 27. It distinguishes search access, app personalization and on-screen content, explaining that their controls do different jobs. The guide also warns that some AI requests may involve cloud processing and describes options for reducing access or disabling the new features.

EFF examines the tension between encrypted messaging and AI services that process conversations in the cloud. Its analysis says trusted execution environments can offer safeguards but do not preserve the same privacy boundary as keeping messages on participants' devices. The authors call for clear user choices and caution against automatically sending private conversations to external AI systems.

Vercel has extended Spend Management to Enterprise Flexible Commitment customers at no extra charge. Teams can set billing-cycle budgets for metered usage drawn from prepaid balances and configure email, webhook or production-deployment responses. A budget alone does not halt usage: deployment pausing must be enabled separately, and it does not stop AI Gateway or v0 consumption.

Vercel's mcp-handler 2.2.0 adds experimental support for WebMCP, a proposed standard that exposes tools to agents inside the browser. Developers explicitly opt tools in and load a script from their MCP endpoint. Calls are forwarded to the MCP server using the signed-in user's identity, allowing authenticated tools without a separate browser-side OAuth flow.

v0 can now install private npm or custom-registry packages using shared Vercel environment variables scoped to Development or Preview. Teams can supply NPM_TOKEN for npm's registry or NPM_RC for custom and multiple registries, bringing internal libraries into their builds. Vercel says these credentials are neither shown to the model nor written to the sandbox filesystem; integration status appears in v0 settings.

Google is expanding its AI & Economy Research Program with additional academic advisers and research leaders. Anu Madgavkar and Daniel Rock will help direct work on adoption, labor markets, enterprise productivity and scientific discovery, while Philippe Aghion and Ajay Agrawal join advisory and visiting roles. The announcement describes a research agenda intended to inform future ATLAS updates, rather than new economic findings.

Google describes two fashion-design collaborations using custom Flow tools ahead of New York Fashion Week. Jane Wade’s styling tool let her combine garments and accessories on digital models before producing additional samples, while Sergio Hudson used simulated runway layouts, lighting and props to plan within his budget. These are specific designer workflows; the article does not quantify general productivity or cost savings.

Cloudflare resolved a dashboard search bug on September 18, 2026 that could freeze the interface when looking up Workers KV, typically on a second search attempt. The company said the problem was confined to dashboard searches: Workers and Workers KV themselves were unaffected. A fix entered monitoring at 08:04 UTC, and the incident was marked resolved at 08:15 UTC.

Vercel reports that TypeSafe AI’s Jev reached nearly 13% of AI Gateway’s paid teams within its first day, more than twice the adoption of any previous model launch on the service. Jev is designed for software decisions, returning typed answers and probabilities instead of prose. The report describes early uptake and leaves the durability of that adoption as an open question.

EFF is urging lawmakers to base AI cybersecurity rules on demonstrated risks and established security practices. Its recommendations include isolating high-risk testing, recording system activity and investigating incidents independently. The organization argues that public incident reports can help improve defenses while cautioning against inflexible requirements that assume today's AI systems and threats will remain unchanged.

Cloudflare reported increased errors affecting Durable Objects and downstream services in the Asia Pacific region during a September 17–18, 2026 incident. The company opened its investigation at 23:37 UTC on September 17. A fix was being monitored by 00:45 UTC the following day, and Cloudflare marked the issue resolved at 00:54 UTC.

Vercel added GLM 5.3 FlashX, a faster serving option for Z.ai’s multimodal coding model, to AI Gateway. Vercel describes inference at roughly 200 tokens per second, targeting interactive applications and repeated agent tool calls where response speed matters. Developers can select the model through supported APIs or coding agents while using the gateway’s existing usage tracking, budgets and routing controls.

Supabase has added service-health checks to Advisors, initially covering error rates for its data API, Auth, Storage and Edge Functions. Results are available in Studio and the Management API, with links for deeper investigation. An unavailable check is distinguished from a completed check with no findings. Database connectivity probes and instance-health checks remain planned additions, rather than features of this first release.

Vercel says its September 17 CLI update can publish eligible static artifacts in under a second by skipping the build step. The feature accepts directories containing up to ten HTML or Markdown files and no more than 5 MB in total. It requires CLI 59.16.0 or later and returns a live URL for qualifying deployments.

Sep 17, 21:51 UTC Resolved - This incident has been resolved. Sep 17, 01:09 UTC Identified - Some Pruna models may not be able to scale up and are therefore either experiencing long queue times or entirely unavailable.

Sep 17, 21:49 UTC Resolved - Between 20:26 and 21:17 UTC on September 17, 2026, GitHub Copilot experienced degradation affecting several GPT models, including GPT-5.6 Luna, GPT-5.6 Terra, GPT-5.6 Sol, GPT-5.3-Codex, and GPT-6 Astra.

The UN System Data Commons brings statistics from UN organizations into a connected platform built on Google’s Data Commons technology. Users can explore data through natural-language queries, thematic browsing and tools for AI research assistants. Google says UN experts validate the datasets, while advising readers to check underlying sources before citing important figures. Broader coverage planned for 2027 remains a future goal.

Vercel's September 17 update allows Turbo build machines to be selected for a single deployment, so a temporary resource increase does not require changing project settings. Developers can request Turbo through a GitHub-bound commit message, the deployment API or the CLI's turbo option. The CLI route requires Vercel CLI 59.20.0 or later.

Cloudflare’s official September 17, 2026 status report identified a network performance problem affecting Ashburn, Virginia, in the United States. The initial 18:29 UTC notice said a fix had been implemented; a monitoring update followed at 18:32 UTC. Cloudflare declared the incident resolved at 19:58 UTC.