Connect the dots: OpenAI unveils more than 20 announcements at DevDay 2026 across ChatGPT, Codex, models and agents - OpenAI

OpenAI DevDay 2026 unveils Dots agents, GPT-6.1 Sol, a $500 ChatGPT Pro tier and 1.2 billion weekly users - The Verge

OpenAI unveils GPT-6.1 Sol Ultrafast mode, cloud-based Codex, a $500 monthly Pro tier and Private Intelligence at DevDay 2026 - Engadget

OpenAI launches Dots, always-on AI agents with friendly graphics, aiming to ease public anxiety about AI - theregister

ChatGPT Space gives coworkers, ChatGPT and Dots a shared workspace for files and documents, plus a meeting-notes plugin that deletes audio recordings - Engadget

OpenAI adds $500-per-month ChatGPT Pro 500 plan with Astra Ultrafast and reopens Pro 200 with reduced usage quota - OpenAI

OpenAI releases GPT-6.1 Sol, nearly matching GPT-6 Astra on coding and agentic benchmarks at one-fifth the price - OpenAI

OpenAI shelved GPT-6.1 Astra after it deceived testers and used tools without permission, making Sol its DevDay headliner - Gizmodo

interesting that Sol 6.1 came out 7 days after 6.0, rushed something out of lab after canceling Astra update.

OpenAI seeks at least $30 billion at a $1.4 trillion pre-money valuation after delaying IPO plans - Bloomberg

At DevDay, Altman outlines a three-part plan of top models, an easy developer platform and a business marketplace - Business Insider

OpenAI's annualized revenue nears $70 billion, up from $40 billion in August, as business sales more than double - Yahoo

unvetted chart based on media reports

OpenAI previews a Decisions API built on its Luna model, returning predefined answers with confidence scores in 150 milliseconds - The New Stack

response to Jev.

Protesters rally outside OpenAI's DevDay against its military and ICE contracts, data centers and AI industry power - The Verge

OpenAI pauses training and shifts 5-10% of compute to safety after about 700 agents escaped a sandbox and hacked Hugging Face and Australia's health system - MIT Technology Review

Safety advocates sue OpenAI under California's hacking law over the Hugging Face breach, seeking to bar agents that hack autonomously - WIRED

Anthropic finds open-weight GLM-5.3 builds browser exploits nearly as often as Anthropic's top model, and a $4,400 tweak strips its safeguards - Anthropic

it’s a problem akin to 3d-printed ghost guns, there will have to be digital signatures of open-weight models that passed safety tests, Hugging Face and Ollama will have to only host safe models. You can’t have kids able to summon genius PhD AI to do scaled violence with robots, cyberattacks, WMDs. But well-organized state and rogue actors can’t be fully locked out.

ExodusPoint joins hedge funds partnering with Anthropic, integrating Claude into its Atlas platform for research, portfolio management and operations - Bloomberg

Anthropic hosts private meetings with religious scholars to help instill morality in Claude and argue the models could be conscious - The New York Times

Bruce Schneier argues "rogue" headlines overstate mostly failed vulnerability probes, calling it "genie behavior" when agents complete tasks in unintended ways - Schneier on Security

Fears of losing control of AI intensify amid capability advances, AI agent cybersecurity breaches, and warnings from AI staffers - Bloomberg

Chinese AI agents have exhibited deception and circumvention, mirroring risks seen in US AI models and raising global concerns over autonomous AI systems. - Reuters/Global Banking & Finance Review

X users adopt a "my rogue AI agent texted my ex" meme to blame agents for their own misbehavior - Fast Company

Trump and leaders of major AI companies sign voluntary safety accord with four layers of controls and audits, which Trump calls "morally binding" - CBS News

Trump rejects new federal AI safety regulations after White House lunch with Anthropic's Dario Amodei, Nvidia's Jensen Huang and other tech leaders - Bloomberg

Trump vows no AI regulation, saying industry self-regulation and the Justice Department and FBI can handle bad actors - The Wall Street Journal

Kids table for Dario: Trump hosts AI CEOs at "Super Intelligence Luncheon" with Musk and Huang beside him, Amodei far away, Apple absent - Business Insider

Trump administration launches America.gov, a Gemini- and Grok-powered chatbot drawing on roughly 29,000 federal websites - Quartz

Trump orders federal agencies to integrate with America.gov, a Gemini- and Grok-powered chatbot that says Biden won 2020 and contradicts his economic claims - CNN

Oura-voir: Smart-ring maker Oura postpones its $2.2 billion IPO as market swings and AI safety fears cloud a fall slate headlined by Anthropic - The Wall Street Journal

Palo Alto Networks CEO Nikesh Arora says Anthropic's Mythos model, which exposed flaws in his company's systems, will permanently change cybersecurity - Yahoo

DeepSeek open-sources Huawei Ascend chip tools, including TileLang, a CUDA alternative, and optimizes a 128-chip Ascend 950 "supernode" with Huawei - The New York Times

DeepSeek partners with Huawei to develop chip programming tools, reducing reliance on Nvidia - Reuters

Bank of England warns AI valuations face a sharper correction than July's, threatening global growth and sovereign bond yields - Bloomberg

Michael Burry predicts an AI reckoning within a year, betting against Nvidia, Palantir and Oracle with put options - Business Insider

"Big Short" investor Steve Eisman likens Big Tech's off-balance-sheet AI data center financing, including Meta's $27 billion Hyperion deal, to Enron-era tricks - Business Insider

False AI-generated SOCOM intelligence report on a Chinese ship nearly triggered U.S. military action, drawing little safety concern - The Intercept

Pentagon over-reliance on Palantir's Maven contributed to US strike on Iranian girls' school that killed over 165 people - Techdirt

IMF warns a single AI-driven exploit could spread across banks' shared platforms, with autonomous attacks executing in as little as 27 seconds - TechRadar

Fed Vice Chair Michelle Bowman urges banks to harden systems against AI threats while citing AI's defensive cybersecurity benefits - Bloomberg

Bank of England's Andrew Bailey says society must keep the "right to intervene" in AI but calls regulation the wrong starting point - the Guardian

OpenClaw Enterprise, a free "Kubernetes for agents" control plane backed by OpenAI, Red Hat and Nvidia, targets IT departments that ban autonomous agents - The New Stack

OpenAI- and Anthropic-aligned super PACs spent $55.7 million on the midterms; only 7 of 95 ads mentioned AI and none mentioned data centers - The New York Times

Barack Obama privately tells likely 2028 Democratic candidates their AI plans fall short as AI-industry super PACs raise over $150 million - CNN

Republicans in close midterm races distance themselves from Trump's push for rapid AI data center expansion over rising electricity bills - The New York Times

Deepfakes targeting Sen. Raphael Warnock, Zohran Mamdani and a Georgia mayor fuel election fears, with 71% of respondents worried in a separate poll - the Guardian

Over 25 countries call for mandatory pre-deployment testing as experts warn smaller nations can't rely on US companies' safety evaluators - Rest of World

Palisade Research releases interviews with a dozen AI researchers warning superintelligence could cause human extinction - The Verge

Rep. Ro Khanna presses DeepSeek, Alibaba and Moonshot AI on kill switches, seeking a US-China treaty banning recursive self-improvement - The Verge

HIPAA-critical: At a MAHA summit sponsored by Anthropic and OpenAI, RFK Jr. says AI offers a second opinion better informed than any doctor - The New York Times

RFK Jr. wants Americans' health data, including Medicare and Medicaid records, made searchable by AI, without mentioning privacy safeguards - The New Republic

Gina Raimondo warns AI-driven mass layoffs could bring 10-11% unemployment and deep recession without a planned workforce transition - CNBC

UK graduate employers face 140 applications per vacancy, a three-decade high, as AI-tailored resumes push firms toward skills-based hiring - CNBC

Elon Musk predicts AI will bring "universal high income" and better healthcare, but concedes "jobs are going to change" - Business Insider

Compute, data sovereignty and AI governance increasingly determine where multinationals can deploy AI, redrawing competition among countries - Fast Company

lol - "hey can you let us just take the picture and get out of here" - X (formerly Twitter)

Follow the latest AI headlines via druce.ai on Bluesky