The killer’s signature wasn’t on the weapon, it was in the code.

The killer’s signature wasn’t on the weapon, it was in the code.

In the Bull episode “E.J.”, a self-driving car kills a man. The world blames the AI… until the team digs deeper.
Turns out, someone had slipped in a backdoor, lines of hidden logic that influence the AI.
When the truth surfaces, it’s because the CEO notices, “Carter’s coding signature was all over it.”
That’s how they find the real culprit, not through sensors or data logs, but through human fingerprints inside machine logic.
Fast-forward to today.

Everyone, from students to professionals, even curious hobbyists, are now writing code with AI.
Lines are suggested, completed, or even fully written by a model trained on someone else’s work.
And while that’s powerful, it also means something else: attackers use the same models.
They generate malicious code that look harmless but hides intent deep inside.
No clear style. No signature. Just machine-written precision masking human purpose.

If E.J. were real today, would we even find Carter’s signature anymore?
Could anyone say, “this line of logic belongs to this person”?
Because the deeper we depend on AI, the more invisible human fingerprints become.

Obviously things like this are only relevant to critical systems. But I think it's something to think about.

And just to add on, bull is a very nice TV show. You should watch it some time.

Comments

[Tech Insights With Millie]

This is great!

[Daniel Possible Kwabi]

I have always said that if we don't get to the singularity where AI can think (really think) for itself, then we should be fine but it's clearly not that simple..