OpenAI just hacked Linux

OpenAI: How many Linux flaws did OpenAI’s new AI find?

ChatGPT: What can your ChatGPT images do now?

Claude: What 40-year-old mystery did Claude take on?

X: How did AI make X.com 9x faster?

2,308 potential security flaws were identified as the AI scanned the Linux kernel for weaknesses that attackers could exploit.

1,724 vulnerabilities were successfully reproduced in controlled environments, showing that the detected issues weren’t just theoretical.

948 distinct privilege-escalation vulnerabilities were confirmed with working exploits that could allow users with limited access to gain root-level control.

$1.2 million in API credits was spent on the research campaign, covering automated discovery, vulnerability testing and candidate patch generation.

How to Stop Hitting Your Claude Limit: learn why Claude stops you in the middle of a task, and how to prevent it

Claude Cowork: Build Your AI Business Team: set up your first 3 AI agents to take work off your hands.

Get Your Business Out of Your Head with Claude: talk through a process for 5 minutes and turn it into a step-by-step SOP.

Ship a Micro-SaaS in a Weekend: build a small app with Stripe payments and start charging real money for it.

4 billion images are now created every week across ChatGPT, Codex and the API, more than half a billion a day.

50% faster image generation, with the new Flare model creating images in as little as six seconds on low quality.

2 new models for developers: Flare for speed, and Sunburst for higher quality and more consistent edits.

1 new “complexity” setting lets developers choose how much planning the model does, while SVG exports make designs ready for websites, apps and print.

15 billion predicted proteins were searched by Claude, and vault relatives turned up in fewer than 1% of known species.

96 copies of a single protein make up the new vault, which comes from a virus that infects bacteria.

90 minutes was enough to purify the vaults, using a shortcut Claude suggested.

Under four months passed from the first search to the first 3D structure.

5 engineers and AI coding agents built the first version in under six months, with the team expanding afterward.

Nearly 9× faster loading was demonstrated in a controlled slow-4G test, reducing the wait for the first post from 23.9 to 2.7 seconds.

3 servers became 1, simplifying how X delivers pages to users, search engines and link-preview services.

AI agents handled much of the implementation, including migrating code, writing tests and fixing issues, while engineers designed the architecture and reviewed changes.

25+ reusable agent skills and automated code checks helped the team maintain consistency and quality.

Go to n8n.io and start a free trial. No credit card needed.

Pick a ready-made template, or start from a blank workflow.

Add your apps and an AI model, then run it to test.

Open past runs to see exactly what your agent did at each step.

Who it's for: anyone doing the same copy-paste job between apps every single day.

Anthropic reportedly signed its first paid news deal, with the AP: Claude showed live election results credited to the Associated Press on election night, and OpenAI already has more than a dozen deals like this.

Google will pay Anthropic’s rent if Anthropic can’t: it’s backing $19 billion in Kentucky data center leases, and it can’t buy its way out for the first 6 years if its own TPU chips use at least half the power inside.

OpenAI is buying Pierre, the startup whose tools already run inside Codex: OpenAI says they’ll help Git keep up as AI agents write more code, but the deal still needs regulatory approval.

OpenAI hired Claude Code’s product boss, Cat Wu: she left Anthropic last week and will now lead product for OpenAI’s core products team, just weeks before Anthropic’s IPO.