AI-generated malware, and a surprising AI breakthrough • Graham Cluley

In episode 44 of The AI Repair, ChatGPT gained’t construct a crystal meth lab, GPT-4o improves the present’s podcast artwork, some college students handle to screw in a lightbulb, Google releases Gemini 2.5 Professional Experimental and no one notices, and Mark invents a clock for measuring AI time.

Graham explains how ChatGPT’s love for Younger Grownup fiction can be utilized to show it into an evil malware developer, and Mark seems on the ARC-AGI-2 benchmark and a staggering leap ahead in AI intelligence.

All this and far more is mentioned within the newest version of “The AI Repair” podcast by Graham Cluley and Mark Stockley.

Powered by RedCircle

Hosts:

Graham Cluley:

 @grahamcluley.com

 @[email protected]

Mark Stockley:

 @ai-fix-mark.bsky.social

Episode hyperlinks:

Assist the present:

You may assist the podcast by telling your folks and colleagues about “The AI Repair”, and leaving us a assessment on Apple Podcasts or Podchaser.

Observe us:

Observe the present on Bluesky, or subscribe without spending a dime in your favorite podcast app akin to Apple Podcasts or Spotify, or go to our web site for extra info.

Discovered this text fascinating? Observe Graham Cluley on Bluesky or Mastodon to learn extra of the unique content material we put up.