When AI stopped helping investigate the attack #Shorts

What if the artificial intelligence that was supposed to help you investigate an attack stopped cooperating just when you needed it most?

That is what OpenCode reported during its response to the intrusion. According to its cofounder, the attackers kept trying to break in for about 48 hours. The team used models to examine malicious code, but some later began refusing the task: the attack material triggered their safety filters.

Investigating a threat requires reading dangerous instructions, even when the goal is defense. OpenCode said access to specialized models for defensive tasks allowed them to continue investigating. The case raises an uncomfortable question: how can we help those analyzing an attack without enabling one?

Full episode: https://youtu.be/wHL-XAA8fII

🤖 AI-generated content: the script, voices, and images in this episode were produced using artificial intelligence tools.

#Shorts