Originally published at: https://boingboing.net/2024/04/26/man-charged-with-framing-maryland-principal-with-ai-generated-racist-rant.html
…
5 Likes
Yikes. But, like, we are, what, 3-4 years away from you not being able to tell the difference.
The future is going to suuuccckkk.
There needs to be like encrypted code or something that can’t be stripped out that signifies something is AI generated or not.
8 Likes
And a skilled sound engineer could make the recording sound like a real phone call right now.
8 Likes
There used to be a meme on slashdot suggesting that any simple technical solution usually ran into problems when applied to social issues.
Most of the issues I see are logistical problems, how do you implement it and ensure it’s used only in AI or not, how do you distribute it, and how can you ensure that all media contains it?
1 Like
It was real-sounding enough to fool a ton of people, and this was created by an athletic director who probably had zero expertise in this area. (He certainly didn’t cover his tracks very well, based on the information that police have released.)
Someone using state-of-the-art technology who is also a little more careful could probably create recordings that would be difficult for even forensic experts to identify as fake. And as we’ve seen it certainly doesn’t need to be perfect to cause a whole lot of harm.
Edit to add:
Even though the main culprit has been caught, it sure sounds like the principal is going to have difficult relations with the rest of the staff, including one teacher who believed the accusation and used a student to help spread the recording on social media.
8 Likes
The only way forward here is that people need to act as though provenance matters for media including photos, video and audio. And that media without provenance, meaning a source willing to put their name and reputation next to it, should be ignored by default.
People do this with text because they are taught to from a young age. People mostly do this with photos, because they were taught to since around 2010 or so.
We need to teach this for audio and video or shit like this will keep happening. The lesson is: the fakes are flawless, so you need to start looking at the source. In this case, it was an anonymous email which maybe should have sent up some red flags.
4 Likes
Not as often as they should. There are still a lot of people who believe The Protocols of the Meetings of the Learned Elders of Zion is a credible document even though it was debunked as a racist forgery over a century ago. And of course Qanon garbage and other lies aren’t exactly in short supply either.
5 Likes
From 2012. I remember hearing about this around that time. The hum of the electrical frequency in the wires, picked up by microphones.
“It is a technique known as Electric Network Frequency (ENF) analysis, and it is helping forensic scientists to separate genuine, unedited recordings from those that have been tampered with.”
I don’t know if it’s use is widespread or confined to audio forensics.
9 Likes
People should definitely be more skeptical about random garbage someone posted on the internet; but things get a bit tricky when it comes to whistleblower material, or stuff that would be if it were real.
There are absolutely things that can be true and very much in the public interest where attaching your name is not something you do for good health and prosperity; so discarding anything without someone willing to admit to taping it raises some problems in those cases.
There’s an issue in the other direction, as well, with people whose business is more or less attaching their names to sensational tidbits and not facing any consequences for accuracy(have Alex Jones’ assets been flushed out yet?). That, at least, can be solved in principle by parsing more closely(‘repeated by a talking head without an attributed source’ isn’t operationally all that different from ‘found on social media without an attributed source’, when the reputational risk to the talking head is so low); but we have good reason to believe that people often don’t parse more closely, and do attach the credibility of the person repeating the story to the story; and sometimes that’s even a correct course of action(actual journalists who do actually care about their credibility and the accuracy of their stories often need to not credit sources if they want access to information that PR isn’t willing to release).
3 Likes
This is where journalism is worth paying for. I have some (I assume fairly rare) experience working with large leaks to media outlets (example). The verification work we did before putting the organization’s credibility behind a document leak is the most intense and delicate work in the field. Because there’s always a reason and it’s usually not journalism so there’s a bit of a dance and a lot of covert effort into outside corroboration.
Agree that the collapse of journalist / pundit / youtuber has made this all much harder. Not everyone is wagering with the same chips.
4 Likes
I get that, but it’s certainly not a perfect solution either. One recent example: We still don’t know who made the secret (and real) recordings of that Los Angeles City Council redistricting meeting where council members were making racist comments. The fallout from that was significant, with two councilmembers eventually resigning, and that wouldn’t have happened if everyone just ignored it because they didn’t know the source.
4 Likes
I remember that in some crime drama (maybe was Columbo, or was it The Professionals?) the plot revolved on a fake cassette recording made up cutting and splicing tape recording of the victims speech. I think that NBC and ITV sound engineers could have been able to make a realistic fake recording easily.
And in more recent time, learnig to use a DAW makes that easier to do, like in the case below, but of course using AI tools makes it easy to do without having to learn to use any tool.
1 Like
This topic was automatically closed after 5 days. New replies are no longer allowed.
