Here are today's news for 20th of November about Google, Anthropic, Microsoft and U.S. senators. Let's dive into the details. Today in brief: Google rolled out Gemini 4 Hydrogen, a faster, lower, cost model for coding and agent tasks, across Search and other AI products; Anthropic launched Claude Meetings in beta, letting Claude join Google Meet calls and help produce notes and work artifacts; Anthropic priced its IPO at $83 per share and began Nasdaq trading on November 20, with its opening price valuing the company at more than $2 trillion; Microsoft introduced seven new AI models at Ignite, including models for cybersecurity, coding, image generation, and transcription and A bipartisan group introduced legislation requiring public companies to disclose major AI, related financing and improve investor warnings about AI risks. On November 12, 2026, Google introduced Gemini 4 Hydrogen, a model designed for fast coding and multi, step agent tasks, and began a global rollout across Google Search, the Gemini app, AI Studio, the Gemini API and business products. In Search, Hydrogen is available to all users and is being used for difficult questions in AI Overviews; developers can also access it through the API, while paid AI Pro and Ultra subscribers can use it in Antigravity. Google says Hydrogen completes tasks 2.5 times faster than Gemini 3.8 Flash and scored 57% on Terminal, Bench 4.1, versus 18.2% for Flash. Its Artificial Analysis Intelligence Index score is 51 at default “high” reasoning, compared with 50 for Gemini 4 Argon, rising to 54 at “xhigh.” The API price is $1.50 per million input tokens and $7.50 per million output tokens, with a one, million, token context window. On Nov. 12, Anthropic began rolling out Claude Meetings in beta to Team and Enterprise subscribers, adding Claude as a visible participant in Google Meet calls. It can answer questions in chat or aloud, use connected tools with permission, and draft documents, whiteboards and other work artifacts as people talk; afterward, notes, transcripts and action items are available in Claude. The shift is from meeting summaries to an AI that helps produce work during the meeting, putting Anthropic in competition for workplace workflows where Microsoft and Google already integrate assistants into collaboration software. Beta notes, transcripts and ordinary in, call answers are free; connected, tool use counts toward plan usage, while generated artifacts are free through Feb. 10, 2027. The rollout excludes some organizations using zero data retention, HIPAA, ready configurations or customer, managed encryption keys. Anthropic priced its IPO at $83 a share on Nov. 18, 2026, then began Nasdaq trading under ticker ANT on Nov. 19, not Nov. 20. It sold 980 million shares, raising about $81.3 billion in the largest IPO on record and setting an initial valuation near $1.5 trillion. Shares opened at $111.50, up 34%, briefly lifting the fully diluted valuation to roughly $2 trillion; they closed at $97.17, 17.1% above the offer price. The debut tests whether investors will fund frontier, AI growth at a scale that intensifies competition with other technology firms while confronting concerns about an AI bubble and the costs of computing infrastructure. Anthropic, founded less than six years earlier, makes the Claude models and coding assistant Claude Code. Its prospectus disclosed $6.2 billion in debt and commitments of about $900 billion for computing and infrastructure in coming years, including $70 billion over the next 12 months, making access to public capital consequential to its expansion. At Ignite 2026, held Nov. 17, 20 in San Francisco and online, Microsoft announced seven AI models spanning coding, cybersecurity, image editing, speech, transcription and decision, making. The lineup comprises MAI Cyber 2 Flash, MAI Code 2, MAI Code 2 Flash, MAI Image 2.7, MAI Realtime 1, Dragon Copilot Transcribe and Decision 1.1. Microsoft says Code 2 scored 98% of GPT, 6.1 Sol’s performance across four coding benchmarks at about 40% of its price; Cyber 2 Flash matched GPT, 6 Astra on CyberGym, E2E for under 2% of the cost per task. Image 2.7 ranked first in Artificial Analysis image editing, while Realtime 1 starts at $0.01 per audio minute. The models target businesses seeking lower, cost alternatives to OpenAI and Anthropic, and deepen Microsoft’s shift from relying on external AI to building specialized models for its own products. Dragon Copilot Transcribe, based on MAI Transcribe 2, already serves 10 million patient encounters monthly, Microsoft says. On Nov. 19, 2026, Sens. Mark Warner, John Kennedy, Angela Alsobrooks and Mike Rounds introduced the bipartisan AUDIT AI Act, directing the SEC to set new disclosure rules for publicly traded companies. The proposal targets nonfinancial firms’ AI, infrastructure joint ventures, special, purpose vehicles and affiliated entities, plus financing commitments such as debt, leases, guarantees and servicing agreements in offering statements and periodic reports. It would also standardize AI, risk warnings, covering reliance on outside models, concentration among computing providers, changing training and operating costs, and data, rights liabilities; companies would disclose whether they have AI, governance policies and board oversight. The SEC would have 180 days after enactment to write financing rules. The bill responds to AI infrastructure spending measured in trillions and opaque arrangements that can hide exposure, including circular investments in which firms finance customers. Warner called the scale a “systemic risk. That's all for today's AI news briefing. Have a great day!