So begins the battle for truth?—?it’s not the fight you’re expecting

The first shot won’t feel like much if it hasn’t happened already. It will be a simple, calculated attack that will trigger a devastating change in the way humans receive information. Most of us won’t even know it has happened. If it hasn’t happened already.

It might be a speech by an elected official, a president, a prime minister, or a chancellor. It might be aired on multiple networks and radio stations, appearing in podcasts and blogs. It could be a speech about economics, war, or a party initiative, yet it will be perceived ever so slightly differently, not because of the spin (to the right or left), but because it will actually be a different speech.

A few words spoken will have been altered. The change undetectable to anyone who wasn’t there, if even by those that were. Imagine cameras and audio recorders in the room capturing, and then passing on to a network for broadcast, a recording of what actually occurred for syndication, and then another, sending out an altered version. It will be something that only a machine would be able to pick up if it could compare it to the real thing, but none the less it would be out there. A slightly different speech from the original, and it would be detectable only if all recorded versions were analyzed and compared against each other at the file level.

The changes would have been made from a computer, possibly a phone, and would be delivered without the subject or the person capturing the recording being aware of it. The amendment invisible, embedded within the media package itself and replicated across the news and media supply chain almost instantaneously. “10,000 wounded and 1,000 missing or presumed dead,” becoming “1,000 wounded and 100 missing or presumed dead.” Or “2% becoming 2.5%,” or “51% of voters becoming 50%.” You get the picture.

Both versions would propagate at the speed of social and digital sharing. Both versions would appear in different newspapers on the same newsstands and on different news shows at the same time. But in the case of merely a false claim in need of a retraction, as we might have today, both versions would be digitally correct.

Both versions were said by the person on the screen, in the audio recording, or written on the page. Both “recorded at the same time” and both could be “proved” as being the one that was captured at that moment and time that it was said.

So the question is: How can you take a person at their word if their words have been altered? How can you say “I never said that” when there video and audio of you saying it? If we thought we had a problem with the twisting or spinning of words or facts, imagine if there were two versions of the same event?

So not only is this alteration of video and audio possible, it’s easy and will only get easier with each passing day.

I first discovered Lyrebird.ai a few months ago and was struck by its ability to create voices. The demo, with the video, for example, of Barack Obama (below) truly shook me. It reminded me of the first time I saw a postage stamp sized video on the internet. Few imagined at the time that these grainy low-res videos would lead to hi-res streaming (and in such a relatively short time). Remember early analog cell phone calls? Or early Skype?

It struck me that if Lyrebird (whose mission I admire a great deal) is offering the creation of voices on their own site in under a minute now, then what will this technology look like in a year or two years time, but with a slightly different mission in mind.

It also struck me that Lyrebird has to (or has chosen to) post the words:

THIS IS AN ARTIFICIAL VIDEO CREATED BY LYREBIRD. IT DOES NOT CONVEY THE OPINION OF BARACK OBAMA.

Now think about the opening I mentioned using this similar, or perhaps nation state-controlled and more advanced versions of this software (which is free today), in an actual speech or press conference? Imagine an hour-long speech, with only a few words or facts altered. As I said, undetectable to most, but with sufficient change in meaning to have an impact. Who would be able to surface the truth (and how)? would you trust? Which version?

Now imagine a hostile state actor wanting to convince their population of something? Think of how a few changes to a speech or the complete fabrication of a new one, could alter sentiment in a crisis, or when, for example, a story is unfolding in real time, and there’s no time to fact check. Imagine the classic, “This just in: we have received of [insert name] saying [insert words].” And you can begin to see where this could go. Imagine a hostile actor going back through digital archives and altering them, not a lot, but enough to where that show or speech is just slightly different from the way you might remember it.

For context, you can create your own digital voice on Lyrebird in 60 seconds.

here’s my the

Lyrebird voice was created from “70 valid recordings.” of small phrases that the app prompts you to enter.

Again, if this today is the equivalent of the postage stamp video during the early days of the internet, then what will the hi-res version look like? (No offense intended team Lyrebird.)

So what can be done?

Today’s media file formats are ancient. To put this into context, all of the music that you listen to, whether it’s on a streaming service or digital platform, is delivered to that platform in a 26-year-old file format called the WAV. As in: “your_song_name.WAV.” Its evolutionary partner is the MP3, which as you know arguably helped (or broke) the music industry, was launched into the world three years later. The two recordings above are both MP3s, for example. What’s interesting is that it is in one of those vessels that the first voice hack will occur, and this is possible because those file formats are not capable of holding any data. They do not track where they were recorded or what the device was that recorded them. To be fair, some amount of information can be written into them, but the issue becomes that it can all be removed, altered, or merely copied and made into something else. In fact, most file formats were designed to hold just the content. The data around it was not really viewed as necessary. But to me this is everything, and this is how we can solve the outlined disaster scenarios that can, and I believe occur if we don’t act.

I published the first of many blogs in November and began with an amazing team to build a new media format that we call dotBC, i.e., “Your_Song_Name.bc” to solve a music-specific problem.

Essentially, each song that you hear has multiple songwriters, performers, and musicians, all of whom are owed money when the song is used, but who have no way to add their information into the song itself. This is because, as I mentioned above, the song files themselves do not contain within them a way to keep or change information on ownership. This has resulted in millions of dollars in lost revenue for artists and songwriters, and hundreds of millions of dollars in lawsuits. In short, it’s a mess.

But by creating a smart container for all media and having an immutable blockchain or shared data ledger track the changes to the song’s owners, over time you can create an audit trail of the song from its creation to distribution and beyond. A song’s creation date and location (and anyone who has a vested interest in it) are all contained within the media file itself, and the blockchain means that the data can all be tracked and corroborated from day one.

As the project evolved, it occurred to me that this file versioning technology would be one of the only ways to accurately track and corroborate different versions of recordings made of public and private figures. And if you tried to enter any false version it would expose itself as such through the authentication and validation system that keeps the blockchain ledger running. The altered version would be proposed, and need to be approved by parties to the media, in which case it could be identified and tracked back to the person or machine that created it. You could literally catch the issue by the timestamp on the blockchain, and through the identity and authority model and protocol that we have been working on, and that will be available to all and open source in 2018.

If we apply the technology that we have been building for music to all media, then the audio and video track could not be altered without its owners knowing it. It could therefore not reach broadcast capability due to filters on identity and ownership that could be applied cryptographically. This would not preclude the software from creating videos or audio like the ones above. It would not alter or affect Lyrebird’s brilliant mission (see their ethics statement with which I 100% concur) but it would allow for them to assist their users in creating, corroborating, and anchoring the identities of their users into a ledger that would prevent these voices being used for fraud or malice.

I believe that there is a narrow window of time to get this right and that it will take a lot of people embracing a new way of working for this to work, but that in the end, this might eclipse even the “fake news” issues that we have seen so much of in recent history. (I will address these in a subsequent blog post on the application for news sources and citations as it relates to news.)

There can be no blockchain of truth to my thinking, but there can be a blockchain of claims, amendments, and corroboration, and so if we start now we can prevent what could become one of the most significant problems we will face in the battle to know what is true.

For more information on the dotBC project please see www.dotblockchainmedia.com

(reposted from Medium)?