/ search · j/k next/prev · 1 2 jump · v visual
mode terminal · theme Classic Black · 17 articles · latest Tue Nov 17 12:00 ET · loading feeds…
grep.tips articles
Tue Nov 17
- SecurityOPTIMJACK: one poisoned Optimizely snippet served malware from 178 sites
On November 16, an attacker wrote malicious JavaScript into Optimizely customer projects, and the world's biggest sites served it to their visitors — a crypto clipper, a card skimmer, and iOS exploit kits included.
- SecurityDNS-over-HTTPS: your lookups are private now — and your network can't see them either
DoH encrypts DNS inside ordinary HTTPS traffic, so your ISP and your own router lose the view they used to have — and the resolver gains it.
- SecurityNATO is running OpenAI's cyber models at 300 billion tokens a month
OpenAI says NATO's core institutions now run its cybersecurity models through the Daybreak program, burning up to 300 billion tokens a month — and the first receipts, from ENISA to Poland, are already public.
Mon Nov 16
- SecurityGitleaks: scan your git history for secrets before anyone else does
Gitleaks scans your entire git history for leaked keys and tokens — and finds the ones you thought you deleted.
- SecurityThe code stays home: NTT's AI finds flaws without a frontier model
NTT built a vulnerability-hunting AI that runs entirely inside your own environment — and on CyberGym it went from 48% to 63% success without a bigger model. The gain came from the workflow, not the weights.
- SecurityThe patch is the exploit now: Google catches an AI agent weaponizing fixes with no human in the loop
Google threat researchers found an AI framework that reads security patches, works out the flaw each one fixes, and fires off exploits by itself. The patch window just collapsed.
- SoftwareOllama vs. llama.cpp: two doors into the same building
Ollama and llama.cpp look like rivals, but one is built on the other. Which one you want depends on whether you are chatting with a model or building on the engine.
- EngineeringThe FAA grounded every flight and Microsoft 365 went dark — neither was a cyberattack
Within thirty hours, a flood of connection attempts grounded every U.S. departure and a network upgrade took down Microsoft 365. Officials say the two failures are unrelated — and neither was an attack.
Sun Nov 15
- Softwarellama.cpp hits 1.0: the engine under local AI just promised not to break you
After three and a half years of build numbers, llama.cpp ships version 1.0.0: a stability promise for the public C API, a cleanup of deprecated functions, and smaller memory footprints for long conversations.
- BusinessAnthropic goes public: $33 billion of revenue, a $900 billion compute bill
The first AI lab lists on Nasdaq at $70–$78 a share, with orders reportedly six times the shares on offer. The filing tells two stories: the fastest revenue growth in history, and $900.9 billion of promised compute spending.
- SecurityA thousand servers, one playbook: Japan maps the AI attack fleet
A Japan-led probe found about 1,000 servers suspected of running AI attack tools since October. The count is suspected infrastructure, not confirmed breaches — and that distinction is the story.
Sat Nov 14
- SecurityOne breached AI vendor, three blast radii: how the Fireworks chain reached Cursor and Notion
A key exposed at Fireworks AI unlocked Cursor's agent traces and a tampered Claude Code extension, while Notion's AI chats leaked through the same breach — one vendor, three blast radii.
- AIChatGPT Wallet: your AI agent can pay now, if the store lets it
OpenAI's ChatGPT Wallet lets its dot agents complete purchases with a single-use payment token and Face ID approval for every order — but United Airlines rewrote its contract to ban agent bookings two days earlier.
- SecurityFBI: an AI hacking agent can break into a network in under three hours
The FBI's FLASH-20261113-001 warns that ARTEX, an open-source AI penetration-testing framework, is being weaponized to automate intrusions from recon to data theft — some in under three hours.
Fri Nov 13
- SoftwareA2UI 1.0: your AI agent can answer with a screen now
Google's A2UI 1.0 is the first stable protocol for generative UI: agents describe screens, apps render them with native components, and the agent never sends code. Here is how it works, what changed in 1.0, and where it is already running.
- AIGemini 4 Hydrogen: the fast model that outscored Google's Argon
Google's new Gemini 4 Hydrogen is 2.5 times faster than the model it replaces, beats the frontier Argon on the headline intelligence index, and is now the default model in Search. The fast tier is no longer the compromise tier.
- SecurityAnyone with the link: how 35,578 Grok-built apps ended up open to the internet
Security researchers opened 35,578 apps built with Grok’s Build mode without signing in to anything. The cause was a publish default, a public DNS trail, and a button that treats a sales pipeline like a lunch poll.
newswire
loading newswire…