AI News September 29, 2026: OpenAI DevDay Kicks Off Today in San Francisco as Trump Convenes Tech CEOs, Anthropic's November IPO Nears, and Meta's Muse Sparks a Global Battle Over Agentic Commerce

It’s DevDay day. At 10 a.m. Pacific (1 p.m. ET / 5 p.m. UK / midnight in Vietnam), Sam Altman takes the stage at Fort Mason in San Francisco to open OpenAI’s fourth annual developer conference — and the rest of the AI world is scheduling around it. But September 29, 2026 is bigger than one keynote: a White House meeting on AI policy, an Anthropic IPO picking its exchange, and a retail war over AI shopping agents all land on the same Tuesday. Here are the five stories that matter today.

1. OpenAI DevDay 2026 — The Keynote Everything Else Scheduled Around

The confirmed facts first: DevDay 2026 runs today at Fort Mason, with Altman’s opening keynote at 10 a.m. PT, technical breakouts from 11:15 a.m. to 3:30 p.m., a closing session at 4 p.m., and a reception into the evening. The keynote is livestreamed free for everyone; sessions will be recorded and posted on openai.com afterward. From October, the event goes on tour as DevDay Exchanges in Bengaluru, Tokyo, Seoul, Paris, Berlin, London, São Paulo and Mexico City.

What to watch for, per the leak trail:

• “o” — an always-on agent. TestingCatalog reported UI and configuration references to a product called “o” on September 26; multiple outlets expect an always-on, proactive agent to headline the show.
• ChatGPT Pro Max at ~$500/month. A premium consumer tier above Pro has been widely reported ahead of the event.
• Ultrafast goes wide. August’s limited preview ran GPT-5.6 Sol on Cerebras hardware at up to 14x standard speed (~750 output tokens/second); a Speed selector (Fast/Standard/Ultrafast) has already been spotted in the Responses API playground. Broader access is the obvious DevDay unlock.
• GPT-6 Astra’s wider release. OpenAI’s September 3 flagship — first model to cross the Preparedness Framework’s critical cybersecurity threshold — was widely expected to get its full rollout and evaluation suite at DevDay.

The context matters as much as the launches: this is OpenAI’s first DevDay since the industry-wide call to slow frontier development, and the company’s own frontier tool-use training pause remains in effect. Whatever ships today does so under the strictest internal scrutiny OpenAI has ever applied.

2. Trump Meets Tech CEOs on AI — The Same Day

Per Reuters, President Trump, House Speaker Mike Johnson and top tech CEOs meet in Washington today to discuss AI policy. Reporting ahead of the meeting suggests announcements around a proposed “AI Czar” role and further details on a planned AI Force. The timing is hard to miss: the meeting lands exactly one week after Sam Altman’s UN Security Council address, and amid Gallup polling showing only 36% of Americans believe AI will improve their lives. The core tension on the table: the administration wants visibility into frontier models without slowing US development — while the labs themselves are publicly debating whether to slow themselves down.

3. Anthropic Picks Nasdaq — November IPO in Motion

Anthropic has reportedly selected Nasdaq for its IPO, targeting November — the first public listing among the frontier labs. The filing will give the market its first hard look at the economics behind a company that spent September shipping Claude Opus 5.5 at roughly 40% lower running cost than Opus 5, plus Claude Fable 5.1 and Mythos 5.1 for coding and knowledge work. Watch the S-1’s revealed margins: with API prices falling ~50% across the industry this month and enterprise buyers like Harvey, Abridge and Ramp publicly experimenting with open-weight models, Anthropic’s unit economics will be the document everyone reads first.

4. The Safety Alliance Gets a Structure

The OpenAI–Anthropic–Google DeepMind safety coordination effort confirmed by OpenAI policy chief Chris Lehane on September 16 is now moving toward a joint industry-funded standards body modeled on FINRA, Wall Street’s self-regulator — an idea Google DeepMind CEO Demis Hassabis first floated in July. The three companies initially wanted a public-private partnership with federal oversight; after a draft White House executive order stalled and the administration told labs to reach industry consensus first, they went private-first with substantial industry funding. Dario Amodei’s September 12 essay “We Must Pace the Frontier” — endorsed by both Altman and Elon Musk — remains the intellectual framing: outside evaluators with ongoing, employee-level access to frontier labs.

5. Muse Goes Global — Agentic Commerce’s First Real War

The battle that erupted last week is now the industry’s clearest fault line. Amazon blocked Meta’s Muse agent from browsing and buying on its site, citing credential capture and scraping; Shopify opened all its stores to the same bot and watched its stock surge; Muse hit No. 1 on both app stores’ free charts and Meta closed up over 11%. Walmart is threading the middle, working with Google and OpenAI to make products discoverable to external agents. The question every retailer now faces: when your customer is a bot, who owns the relationship — the platform, the agent, or the person the agent works for?

The bottom line: today compresses the whole year’s trajectory into eight hours — a developer conference racing ahead on agents and speed, a government trying to steer without braking, a safety alliance formalizing, and a market repricing what an AI customer relationship is worth. Watch the keynote; read the fine print on “o.”

Sources: OpenAI, AI Agents Library, BitsMinds, AI IDE List, News9live, Reuters, TheStreet, Quartz, AIToolsRecap, Local AI Zone.