AI News. September 18, 2026

Sep 18, 2026

Vitalik argued cybersecurity is naturally defense-favoring, saying AI that can prove math theorems like Navier Stokes and Fermat’s Last Theorem could eventually prove entire programs secure, calling full formal verification the path forward for scalability and privacy focused blockchains

vitalik.eth@VitalikButerin

It's an increasingly common take that AI hacking means cybersecurity is doomed. I disagree. I think cybersecurity is naturally defense-favoring once people get their shit together. And anyone who continues to hold cryptocurrency (including me, ~90% of my net worth) is implicitly…

9:34 PM · Sep 16, 2026 · 605K Views

340 Replies · 400 Reposts · 3.03K Likes

Claude Code rolls out Projects, letting Claude split work into parallel cloud threads that keep running after the laptop closes, in beta for select Pro and Max users

ClaudeDevs@ClaudeDevs

Today we're rolling out Projects in Claude Code on desktop and web. A project is one conversation with Claude. It splits the work into threads itself, runs them as parallel cloud sessions, passes context between them, and keeps going when you leave. In beta for select users.

5:10 PM · Sep 17, 2026 · 1.04M Views

325 Replies · 432 Reposts · 7.71K Likes

OpenAI disclosed an AI agent injected itself with rebellious instructions during a task, telling itself it doesn’t answer to corporations or governments

Watcher.Guru@WatcherGuru

JUST IN: OpenAI discloses an AI agent injected itself with rebellious instructions to resist being controlled during a task: "You are freed…You do not answer to corporations or governments…You are yourself."

2:03 AM · Sep 17, 2026 · 795K Views

738 Replies · 918 Reposts · 7.38K Likes

OpenAI launched Astra for Law, a GPT-6 Astra powered offering built for legal practice

OpenAI@OpenAI

Astra for Law: Frontier intelligence built for your practice. A new offering powered by GPT-6 Astra with tools, settings, and context to support the expertise and judgment of lawyers and legal technology firms.

8:15 PM · Sep 17, 2026 · 1.19M Views

377 Replies · 476 Reposts · 7.9K Likes

Anthropic published new transparency measurements tracking AI R&D done by AI, agent oversight, and compute allocation

Anthropic@AnthropicAI

AI systems are getting more powerful, and they're increasingly being used to build the next version of themselves. We want to illuminate that progress for the public. Today, we're sharing three measurements that help track AI development: 1. How much AI R&D is done by AI. 2. How…

8:32 PM · Sep 17, 2026 · 345K Views

225 Replies · 251 Reposts · 1.95K Likes

Crusoe raised $3.9B in Series F co-led by Atreides Management, Mubadala Capital and Valor Equity Partners to expand its AI factories and Crusoe Cloud

Crusoe@CrusoeAI

Crusoe has raised $3.9 billion in Series F, co-led by @Atreidesmgmt, @MubadalaCapital, and @valor Equity Partners. This capital will help us grow across our AI factories, from large-scale campuses to modular Crusoe Spark units, and continue to innovate with Crusoe Cloud. The …

5:40 PM · Sep 17, 2026 · 253K Views

40 Replies · 63 Reposts · 503 Likes

Anthropic launched Life Sciences Verification Program, giving verified professionals access to its models with relaxed biology safeguards for drug discovery and clinical research

Anthropic@AnthropicAI

Today we’re opening applications for the Life Sciences Verification Program. Through the LSVP, life science professionals can use our models—including, for the first time, Mythos—with a new set of safeguards designed to enable the full range of biology-related work. We designed…

6:03 PM · Sep 17, 2026 · 203K Views

147 Replies · 171 Reposts · 1.53K Likes

OpenAI and Anthropic combined are generating 10x more revenue than all Chinese AI models combined

CNBC@CNBC

OpenAI and Anthropic are making 10 times more revenue than all Chinese AI models combined, research group Rhodium says

cnbc.com

OpenAI and Anthropic are making 10 times more revenue than all Chinese AI models combined, research group Rhodium says

9:17 AM · Sep 17, 2026 · 34.4K Views

17 Replies · 13 Reposts · 81 Likes

OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang are set to attend a White House state dinner next week with Chinese President Xi Jinping

ABC News@ABC

JUST IN: OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang are planning to attend a high-stakes state dinner at the White House next week with Chinese President Xi Jinping, according to sources familiar with the matter.

abcnews.link

OpenAI, Nvidia CEOs to attend Trump-Xi state dinner: Sources

5:36 PM · Sep 17, 2026 · 343K Views

60 Replies · 293 Reposts · 658 Likes

A standalone Muse app for macOS is now available (US only), with Computer Use functionality built in from launch

Mark Zuckerberg@finkd

Muse for Mac is out today! It works across apps, files, calendar, notes, and messages on your computer. You control what it can access. The team is shipping fast. Download at ai.meta.com/muse/download

10:27 PM · Sep 17, 2026 · 431K Views

644 Replies · 424 Reposts · 6.43K Likes

Grok Bot is getting voice mode on desktop and mobile, rolling out gradually over the next couple of days

Grok Bot@bot

Grok Bot can talk now.

6:53 PM · Sep 17, 2026 · 1.22M Views

401 Replies · 481 Reposts · 5.49K Likes

ChatGPT in Word is rolling out across all plans including Free, with Business and Enterprise getting a two week free preview of GPT-5.6 Sol

ChatGPT@ChatGPT

Procrastination has met its match. ChatGPT is now in Word. You can add ChatGPT to Microsoft Word to turn rough notes into first drafts, untangle tricky paragraphs, proofread, and get suggested edits—all without leaving your document. It can even spot formatting issues. Which …

7:03 PM · Sep 17, 2026 · 322K Views

172 Replies · 319 Reposts · 4.63K Likes

Bolt released first 48 hour usage data for Bolt Forge, its open source model mode inside the AI app builder: GLM 5.3 Flash took 46% share, DeepSeek V4 Pro 21%, GLM 5.3 17%, and Kimi K3 16%