AI & Tech · Week 45, 2026

AI/Tech this week: OpenAI finds 948 Linux kernel bugs, $6.3B Binance heist, Meta's scam-ad revenue

Published 9 Nov 2026 · 5 things · about 2 min read · Reviewed by Gulsar, editor

An AI system went hunting for security holes in the code that runs most of the internet, North Korea pulled off a record crypto theft through a fake job interview, and Tesla called time on the Cybertruck.

1. 1Research

   OpenAI says an automated system found and patched 948 Linux kernel security flaws

   OpenAI described an autonomous pipeline that searched the Linux kernel for local privilege-escalation bugs, checked that they were real, and produced patches for them. The company says the system found, validated and generated fixes for 948 such vulnerabilities.

   Why it matters: Linux runs most servers, phones and cloud services, so AI that can find and fix its security holes at scale affects software almost everyone relies on.

   Learn the basics: Local privilege escalation →

   Source: openai.com

2. 2Other

   Binance says a fake job interview led to the $6.3 billion crypto theft

   Binance's incident report says a North Korean crew posed as recruiters, interviewed one of its engineers and had him download a malware-laced coding "assignment". The malware spread to colleagues' machines and internal AI tooling, ending in the October 29 theft. The FBI blamed the group TraderTraitor.

   Why it matters: The break-in started with an ordinary job application, a reminder that attackers now target people and their work laptops rather than firewalls.

   What's next: Binance has hired independent security experts to review the breach and is working with law enforcement to trace and freeze the stolen assets.

   Learn the basics: Supply-chain attack →

   Source: theregister.com

3. 3Policy

   Internal Meta documents point to $16 billion in revenue linked to scam ads

   Reuters reported on internal Meta documents estimating roughly $16 billion, about 10% of 2024 revenue, would come from ads pushing scams and banned products. Investor Mark Cuban says Facebook keeps running deepfake ads using his face. Meta says the documents give a selective view and that it removed over 134 million scam ads in 2025.

   Why it matters: AI-made fake celebrity ads are now a mainstream fraud channel, and older relatives are the most likely to lose money to them.

   What's next: The same documents note the US Securities and Exchange Commission is looking into Meta over advertising tied to financial fraud.

   Learn the basics: Spot AI scams →

   Source: finance.yahoo.com

4. 4Models

   Google's Gemini Omni 1.2 Flash makes 15-second AI videos and cuts the price 30%

   Google released Gemini Omni 1.2 Flash, which generates video clips of up to 15 seconds, up from 10, and lets users tweak a clip through follow-up instructions instead of regenerating it. The price drops 30% to $0.07 per second of 720p video. It is available through the Gemini API, Google Vids and Gemini for business.

   Why it matters: Longer, cheaper, editable AI video brings ad-style clips within reach of small teams that could never afford a film shoot.

   Learn the basics: How AI makes images →

   Sources: blog.google, deepmind.google

5. 5Gadgets

   Tesla will wind down Cybertruck production to make room for its robotaxi

   Elon Musk said on X that Tesla will stop making the Cybertruck at Gigafactory Texas around the middle of next year to free up capacity for the Cybercab. The pickup sold fewer than 50,000 units in total by March last year and went through a string of recalls.

   Why it matters: Tesla is shifting from selling an unusual consumer vehicle to betting on self-driving taxis, which Musk frames as the company's future.

   What's next: Tesla plans to ramp up Cybercab production; the robotaxi currently operates only in Austin with a fleet said to number in the hundreds.

   Learn the basics: AI and robots →

   Sources: wired.com, nbcnews.com