AI This Week: Claude Joins Your Meetings, OpenAI's GPT-6 Cyber and Gemini for Healthcare

This week's AI news was about agents moving into everyday work. Anthropic put Claude into video calls and onto factory floors, Google launched an agent built for hospitals and health insurers, and OpenAI released a cybersecurity model so capable it is only available to vetted security teams. Outside the labs, governments and voters pushed back: France rallied world leaders behind an AI safety charter, New York lawmakers proposed fines of up to $3 million a day, and US voters overwhelmingly backed limits on new data centers. Here is what happened between November 6 and November 13, 2026, and why it matters.

AI News This Week at a Glance

Anthropic Puts Claude in Meetings and on the Factory Floor

Claude Meetings: an AI participant in your video calls

On November 10, Anthropic launched Claude Meetings in beta for Team and Enterprise plans, starting with Google Meet. Claude joins as a visible participant, follows the conversation and works on tasks while you talk: drafting the plan being discussed, mapping a timeline on a whiteboard and answering questions in the chat using your connected work apps. Notes and action items land in claude.ai afterwards. Anthropic built in some guardrails: Claude announces itself in the chat, will not join calls with outside guests unless an admin allows it, and leaves automatically if an external guest joins. The company also opened Claude Whiteboards and Claude Sheets in beta on paid plans and made Claude Data generally available, according to TechCrunch.

Claude for Manufacturing: 17 skills for quotes, quality and maintenance

Two days later, Anthropic released Claude for Manufacturing, a plugin with 17 skills that work from a factory's existing records. Examples include drafting customer quotes from past jobs, writing nonconformance and 8D corrective-action reports after a failed inspection, ranking likely causes of a machine breakdown from sensor data and repair notes, and building a morning production report from overnight output and scrap figures. Connectors cover Autodesk Fusion, Infor, Microsoft 365 and Oracle NetSuite, and every output is routed to an employee for review. It is available to Team and Enterprise customers.

Why it matters: Anthropic is betting that the next growth phase comes from embedding Claude in specific workplaces rather than winning the general chatbot race. For a broader comparison of the two leading assistants, see our ChatGPT vs Claude comparison.

OpenAI Releases GPT-6 Cyber, Images 2.6 and Retires Old Voice Modes

OpenAI's biggest release of the week was one most people cannot use. GPT-6 Cyber, launched on November 5, is a model for finding and validating software vulnerabilities, available through the API and Codex only to approved customers in OpenAI's Daybreak Red security program. According to OpenAI's release notes, it scores 48.5% on the ExploitGym benchmark, up from 32.1% for GPT-5.6 Cyber, and OpenAI rates it Critical under its Preparedness Framework, its highest risk level. The older gpt-5.5-cyber-preview model will be removed from the API on February 1, 2027.

• Images 2.6 (November 4): image generation about 50% faster, GIF creation and saved reference images in ChatGPT, plus two new API models.
• Voice changes: ChatGPT's Advanced and Standard Voice modes retire on December 11, 2026, with the newer GPT-Live handling nearly all voice use. Live video and screen sharing stay on Advanced Voice for now.
• Businesses in ChatGPT (beta, November 10): US companies can claim a business profile, see how often ChatGPT mentions them and, if eligible, build their own Business Agents.

Why it matters: restricting GPT-6 Cyber to vetted teams follows the review-before-release pattern that has spread across frontier labs this year, where the most dangerous capabilities ship to defenders first. The Businesses in ChatGPT feature, meanwhile, signals that being visible inside AI answers is becoming as important for companies as ranking in search.

Google Takes Gemini Into Hospitals and Onto More Pixel Phones

Gemini for Healthcare

Google Cloud launched Gemini for Healthcare in preview on November 10. It is an agent system for hospitals and health insurers that handles multi-step administrative work such as summarising medical records, preparing for appointments, checking coverage and automating prior authorizations. It connects read-only to Epic and Oracle Health electronic health records and can route tasks between models from Google, Anthropic and OpenAI. Launch customers include HCA Healthcare, Northwell Health, Optum, Highmark Health and Stanford School of Medicine. Google says the platform supports HIPAA compliance and that customer data is not used to train its models.

November Pixel Drop: Gemini Intelligence reaches the Pixel 10

Google's November Pixel Drop, rolling out from November 10, brings Gemini Intelligence, first shown on the Pixel 11, to the Pixel 10 family, adding proactive suggestions and custom widgets. On the Pixel 11 Pro and Pixel 11, Gemini can now screen unknown callers in the US with a live transcript. Google also launched Pixel Tag, an item tracker using the Find Hub network and Ultra Wideband, at $29.99 or $99.99 for a four-pack from November 11.

AI Regulation: France's Charter, New York's Fines and the Data Center Backlash

Macron pushes pre-audits for AI models

French President Emmanuel Macron said AI should be banned wherever it threatens children's mental health, ahead of adopting an AI charter at the Paris Peace Forum. The charter asks labs to accept independent evaluations of generative AI systems and the principle of pre-audit and pre-certification before models reach children. Around 50 leaders and executives joined, including the CEOs of OpenAI, Anthropic, Mistral and Microsoft, according to Euronews. EU AI ministers also meet in Dublin this week to review safeguards such as the AI Act.

New York's RAISE Act 2.0 proposes fines of up to $3 million a day

New York lawmakers proposed a tougher version of the state's AI safety law. As reported by PYMNTS, RAISE Act 2.0 would fine frontier AI developers up to $1 million a day for a first safety violation and up to $3 million a day for repeat violations, and would require annual outside safety audits and quarterly reports. It would apply to the most advanced model makers and to any developer that spent more than $1 billion on its models over three years. It is a proposal, not law.

Voters back limits on AI data centers

In last week's US elections, local ballot measures restricting data centers passed in 40 of 41 cases across 13 states, according to Consortium News. Voters in Edgewater, Florida, approved a ban on new data centers with about 77% of the vote, and 11 Ohio municipalities passed outright bans. Data centers are the backbone of AI, and their demand for power and memory chips is already raising costs elsewhere, as our explainer on why phones are getting more expensive in 2026 shows.

Also This Week in AI

• Google DeepMind released AlphaGenome 2, which predicts how genetic variants affect specific cell types, through a non-commercial API.
• AWS published a free local emulator covering more than 200 AWS services that it says was written by autonomous agents.
• United Airlines said it will restrict AI agents' access to its booking platforms, citing credential security and unauthorized transactions.
• J.Crew launched Style Advisor, an AI shopping assistant built on Anthropic's Claude models.
• A Commonwealth Fund study at NYC Health + Hospitals found AI-drafted patient summaries left out A1c blood sugar results for nearly a quarter of patients.

Key AI News Takeaways: Week of November 13, 2026

This page is updated every week with the most important AI news. If you are choosing between the assistants mentioned here, start with our Google Gemini vs ChatGPT comparison.