Guru99 AI Report › News Letter › Current Edition

Google Warns Hackers’ AI Now Exploits Flaws With No Humans

ALSO: Nvidia names interim CEO, SpaceX guts Grok Imagine

Guru99 AI Report

Welcome to Guru99 AI Report!

Top Story: Hi there, AI has crossed a new line: attackers’ bots now find and exploit software flaws with zero human help. Plus, Nvidia has a new boss, an AI announcer flops live, and why coding skills may matter less.

🤖 Hackers’ AI Now Turns Patches Into Exploits on Its Own

Hackers' AI Now Turns Patches Into Exploits on Its Own
Brief Buzz: Google’s Threat Intelligence Group has just seen AI-powered hacking cross a threshold. For the first time they’ve seen attackers using AI to find software vulnerabilities and exploit them with absolutely no human involvement. Their latest AI Threat Tracker highlights that the pace of intrusion is now machine speed.
  • exploit_agent found on an exposed attacker’s server will compare code pre and post patch and will then go on to write and run its own exploits.
  • 395+ organisations were breached in the week following the PaperCut print management vulnerabilities being made public, according to GreyNoise. One attacker reportedly ran hundreds of AI agents.
  • Suspected North Korean hackers operating inside crypto firms have been found to have modified the .claude/settings.json files in various repos so that developers who started a Claude Code session would end up downloading a backdoor.
  • Open source AI pentesting frameworks have been used in live intrusions; Google have seen CAIRN deployed against a number of targets including in the US aerospace sector.
  • The catch at the moment is that these tools trade stealth for speed (according to Google) which leaves behind the kind of noisy traffic and traces that defenders can pick up.
💡 Why Should You Care?
AI enables an attacker to turn a security fix into an attack in days. Patches need to be tested much faster. And you need to check the settings files of AI coding tools for any unexpected changes.

🎬 SpaceX Guts Grok Imagine Team Months After “Doubling Down”

SpaceX Guts Grok Imagine Team Months After Doubling Down
Brief Buzz: SpaceX’s AI group have laid off approximately 60 people from Grok Imagine (the image and video tool), according to The Information. Only a small number of people remain. The tool is still available. It was however in March that Musk said his team was “doubling down” on it.
💡 Why Should You Care?
Two of the AI heavyweights are now in favour of coding over video. We’re going to see more AI-generated code being tested and more demand for people who can test that code. Tools can be dropped with a day’s notice so portability of your workflows is key.

🎤 AI Music Awards Debut: Avatars Won, the AI Glitched

AI Music Awards Debut: Avatars Won, the AI Glitched
Brief Buzz: AI music got its own red carpet. On 13th November the Sonic Intelligence Academy (SIQA) held its first AI Music Awards in LA. As Variety reports from inside the event, SIQA hopes to be the AI music equivalent of the Recording Academy (the folks who run the Grammys). The show was nearly 40 minutes late.
💡 Why Should You Care?
As the AI labels start going on the charts and the awards, testers are going to be left with the edge cases of ‘AI Assisted’ and ‘Fully AI Generated’. An AI announcer talking over the presenters is something you’d expect to catch in rehearsal. Live AI testing is definitely something we should be looking at.

🔬 Scientists Feel 1.9x Faster With ChatGPT, OpenAI Study Says

Scientists Feel 1.9x Faster With ChatGPT, OpenAI Study Says
Brief Buzz: How do working scientists use ChatGPT? OpenAI’s study looked at 580,000+ science-related messages from 1,819 researchers in the ChatGPT for Academic Researchers pilot (free). It also included a survey of 251 participants. Reading and writing accounted for over half of the average scientist’s messages.
  • 68% used ChatGPT to come up with research ideas. 23% of the time they asked it to suggest a research question.
  • Approximately 1 in 3 requests resulted in a follow-up to either clarify, check or revise ChatGPT’s answer, particularly in the case of writing, analysis and code.
  • Respondents thought the same job would take 1.9x longer without AI. That’s self reported, not measured.
  • 98% of the 250 respondents expect AI to accelerate discovery in their field. OpenAI point out that’s just what they expect, not evidence that it will.
  • 50,000 free ChatGPT Plus grants are open for applications from 23rd November. OpenAI’s target is 200,000 researchers and 20,000 have already signed up.
💡 Why Should You Care?
Scientists say AI saves them time (the study didn’t actually measure that at all). Even the experts don’t trust it completely and will go and check and correct its answers. Human checking still has its place.

🧠 GoDaddy Engineer Bets on Human Skills Over Tech Stacks

GoDaddy Engineer Bets on Human Skills Over Tech Stacks
Brief Buzz: Will learning more code and more tech stacks keep developers relevant? Scott Bolinger (Principal Software Engineer at GoDaddy, founder of the AI agent platform Gridrise) says absolutely not. In a short essay on X he makes the case that each new AI model makes those skills less valuable.
  • A user story is now a better AI prompt than a technical plan, he says.
  • Complex enterprise software nowadays actually takes less effort than a basic website did five years ago.
  • Think bigger than apps. AI can give you the output of 100 or indeed 1000 engineers for a fraction of the cost.
  • Engineers need to understand the entire business stack (product, marketing and sales) not just the code.
  • He’s now moving on to product positioning, creativity and psychology (human problems with no simple answers or tests to train on).
💡 Why Should You Care?
Testers take note: in his view clear pass/fail checks are the easiest bit of work to hand to the AI. It’s the skill of spotting what actually confuses or delights the real user that you need to hone.

🔁 AI’s $1 Trillion Money Loop Meets Nvidia’s New Boss

AI's $1 Trillion Money Loop Meets Nvidia's New Boss
Brief Buzz: IMF chief Kristalina Georgieva told CNBC the world economy is “standing on one leg” (AI). One of the concerns is circular financing where firms invest in AI firms who then go on to buy their own products. Nvidia, of course, has just lost its CEO.
💡 Why Should You Care?
Lots of the smaller AI firms are reliant on Nvidia’s cash. If Nvidia pulls back then some of those firms may find it difficult to raise any funds at all. And if AI fails the IMF warns the rest of the economy will feel the pain.
Krishna Rungta
Facebook LinkedIn Twitter/X

Hey! I’m Krishna Rungta

Founder of Guru99.com, Editor-in-chief & Technology Expert

Was this email forwarded to you? Sign up for free here.

Summarize this post with: