IIt sounds too fanciful and outrageous to be true, but nothing is too outrageous for the world the tech bros have left us. The BBC has revealed that several websites and YouTube channels are using AI to clone David Attenborough’s voice and make him say things – about Russia, about the US election – that he would certainly never say.

It’s not the first time this has happened to a celebrity: Scarlett Johansson refused to license her voice to ChatGPT, accusing them of creating the voice anyway, in a character called Sky. ChatGPT’s developer, OpenAI, said Sky was “another professional actress using her own natural speaking voice”, but he pulled the voice “out of respect for Ms Johansson”. Elsewhere, lawyers continue to struggle, using precedents that predate AI’s existence by decades, which is to say, with one hand tied behind their back.

The Attenborough hoax, however, is a different category of sinister. Johansson is a lot of things, all of them excellent, but she is not the global authority on important things that are true. Attenborough may not be the last true embodiment of trust in a compromised world, but I only back away from that statement because I fear it is aimed at Britain. I stand by this: if you can’t hear and believe His voice, then you can’t hear or believe anything.

There were warning signs surrounding AI voice generation about the potential for scams – the risk that it could soon scrape your child’s voice from their TikTok output, create a credible hostage tape and cause you to blow your bank account into the ether before you realized it said the child was in his bedroom the whole time.

It seemed plausible in the sense that people will do all kinds of stupid things when they hear that a loved one is in danger, but it was unlikely – to me at least – because no matter how believable the voice, I think AI would fall. doubts the credibility of the text. It makes it sound terribly human, even when it’s simply trying to describe a sunset. The idea that it could imitate the tone of one of my children seems absurd; I smelled a rat at the words “Hi Mom” ​​(way too much preamble) and laughed out loud when one of them said “please” even supposedly with a gun to his head.

And yet everything, everywhere, knows you better than you think. My Instagram “for you” feed consists entirely of fanatic guys trying to do home workouts while pit bulls get in their way. I never even conversation about all those interests! Apparently, though, it wasn’t the work of Hercules to guess, just as it wouldn’t be that hard to guess that a teenager is sarcastic and has no manners.

Even if this fear of hoax kidnapping is revived, it is still nothing compared to the naturalist’s forgery. Technology divorced from values ​​destroys those values, but with or without values ​​it can also destroy itself.

Forensic DNA is a case study. Detection has gotten so good that if you shake someone’s hand, then grab a coffee cup and drive 200 miles to a gas station where a robbery-murder is taking place, the person who shook your hand can be placed on the spot. The accuracy, far from making DNA more useful, has completely destroyed it; we are back in a world where only an alibi is enough, i.e. the 19th century. (I have that nugget of a true crime podcast, rather than a police force, so it may not be universally true yet. I don’t want anyone to panic and start a minute-by-minute diary of where they are. )

But if AI voice generation gets good enough to destroy trust in the most trusted, it will quickly destroy trust in everything except a voice you can see coming out of a human, which takes us back many centuries to a world where we believed a limited number of people, family members, a few vetted employees and – on a very good day, if we had met him before – a messenger. In fake Attenborough, the scam of all scams, we have been casually ambushed by modern communications.