Attacking without INTELLIGENCE

GPT3.

An Artificial Intelligence.

A general-purpose LANGUAGE GENERATOR…

…or a LANGUAGE PREDICTION MODEL…

…using 175 BILLION PARAMETERS…

…THAT when introduced in May 2020…

…was the LARGEST ARTIFICIAL NEURAL NETWORK ever created.

It was trained by consuming most of the Internet.

45TB of TEXT DATA.

UNSUPERVISED.

GTP-3 was used by Georgetown researches…

… to write misleading tweets about climate change and foreign affairs…

…and PEOPLE… found the tweets PERSUASIVE.

THIS… is DISINFORMATION.

Made easy. Using AI.

The GREATEST trick AI ever pulled, was convincing the world it EXISTS.

This is not INTELLIGENCE.

BUT, as it matures, this type of technology opens up…

…ONE attack vector in PARTICULAR.

That will CHANGE THE GAME for ATTACKER and VICTIMS.

Imagine having the capability to generate HUMAN-LIKE TEXT…

…that adheres to a specific person’s style of writing….

… by providing JUST ONE sentence… of HUMAN-LIKE input.

A COMMAND!

You could send targeted scam mail, texts or chat messages. ATTACKS. AT SCALE.

And IMAGINE… if they had CONSUMED most of the DATA on the INTERNET…

…and they could IMITATE YOU!

REALLY WELL.

Imagine the implications of PEOPLE not being able to tell the DIFFERENCE…

…in between an ARTICLE written by a REAL journalist, vs…

… an Artificial Intelligence. AI.

Think…large-scale FAKE NEWS…

…from ONE LINE… of INPUT.

And imagine what HAPPENS when that is combined with TEXT-TO-SPEECH.

And DEEP FAKES. Like this TOM CRUISE DEEP FAKE.

Its VISCERAL right?

Imagine a FACETIME CALL…

… from someone that uses the SAME WORDS as someone YOU KNOW…

… talks like someone YOU KNOW… AND…

… looks like someone YOU KNOW.

In REAL-TIME!

Our attack surface, as PEOPLE, expands.

The attacks against the HUMAN interface OPEN UP.

And TRUST over anything… or anyone…

… across a DIGITAL MEDIUM…

… will be LOST!

That FACETIME call?!

Against THE SENSES…

I don’t think we are THAT far away from it becoming a REALITY.