Spotify vs. AI | Matt Connarton Unleashed

0:00 Speaker 1: We have some music news. This is interesting.

0:02 Speaker 2: So we've talked about Spotify before. This is interesting to

0:05 Speaker 2: me for a couple of reasons, but you had sent

0:07 Speaker 2: me this. This is from the Hollywood Reporter. Spotify removes

0:11 Speaker 2: seventy five million spammy in quotes, spammy songs, cracks down

0:18 Speaker 2: on AI use by bad actors. The music streaming giant

0:21 Speaker 2: is looking to modernize, I'm sorry, moderate rather AI generated

0:26 Speaker 2: music as it floods its platform. I can see how

0:29 Speaker 2: that would be a problem. By the way, before we

0:32 Speaker 2: get into the story in a little bit of depth.

0:35 Speaker 2: Here again, this is from the Hollywood Reporter. This just

0:37 Speaker 2: went up two days ago, so this is pretty fresh.

0:41 Speaker 2: But not only not only is this obviously, I mean,

0:45 Speaker 2: I can see why this would be a problem, right.

0:47 Speaker 2: AI makes it so easy, like the Puno app, for example,

0:50 Speaker 2: we've used it on the show. You can create music,

0:52 Speaker 2: you know, music that sounds like it was created by

0:54 Speaker 2: real people, really realistic music.

0:56 Speaker 3: Because they're sailing it from real people, but not realistically.

0:59 Speaker 1: But not only that, so obviously this would be a problem.

1:01 Speaker 2: Right. If it's that easy to create it, it's easy

1:03 Speaker 2: to you know, submit it to Spotify or any other

1:07 Speaker 2: streaming platform, right, and then and then they will use it.

1:10 Speaker 2: But the other thing that's interesting to me, and this

1:12 Speaker 2: came up a few years ago Spotify because we had

1:15 Speaker 2: talked about this on the show. This was yeah, this

1:17 Speaker 2: was at least two or three years ago. I think

1:19 Speaker 2: there was a story about how Spotify was having to

1:21 Speaker 2: go through and remove a lot of lower quality, you know,

1:25 Speaker 2: low quality in terms of production value, like stuff that

1:28 Speaker 2: sounded like demos, for example, not actually fully produced mixed

1:32 Speaker 2: and mastered music because people were uploading so much music

1:37 Speaker 2: to Spotify and a lot of it wasn't even necessarily

1:42 Speaker 2: you know, just just the quality like anything that you

1:46 Speaker 2: hear on Spotify should it should at minimum be the

1:49 Speaker 2: same quality that you might hear on the radio or

1:51 Speaker 2: that you might buy in a store, right, right. So

1:53 Speaker 2: that was becoming a problem. So Spotify was having to

1:55 Speaker 2: go through and remove a lot of that. And what

1:58 Speaker 2: was interesting about that is, you know, people don't necessarily

2:01 Speaker 2: realize these streaming services. It's not like, think of how

2:06 Speaker 2: much music gets uploaded to these streaming services every day

2:09 Speaker 2: across the world. But it's not as though they have

2:13 Speaker 2: unlimited storage. True, you know, there there is I mean,

2:17 Speaker 2: you know, they can always make more right. But it's

2:20 Speaker 2: not as simple as like just you can just upload

2:24 Speaker 2: millions and millions of songs all at once and it's

2:27 Speaker 2: just fine. It's not so they have to as there.

2:31 Speaker 2: And look, I'm not please. I'm not expressing any sympathy

2:35 Speaker 2: for Spotify because I you know, artists can plain often

2:39 Speaker 2: and I'll always advocate on the side of artists that

2:41 Speaker 2: Spotify pays out painfully low amounts of you know, for

2:46 Speaker 2: the music that you know, pays low amounts of these

2:49 Speaker 2: artists for making their music available to stream. But it's

2:54 Speaker 2: it's but but these are problems, you know, it's not

2:57 Speaker 2: a perfect business model in the sense that there are

2:59 Speaker 2: pro problems with you know, in terms of storage and

3:02 Speaker 2: and now they're having to go through and deal with

3:04 Speaker 2: the onslaught of AI and release all these and remove

3:08 Speaker 2: rather all these spammy songs that are produced produced by

3:12 Speaker 2: AI and and not only that, but there's also going

3:17 Speaker 2: to be the effect of some people and I remember

3:21 Speaker 2: we talked about this on the show once before. Some

3:22 Speaker 2: people are going to get caught in that net inevitably

3:25 Speaker 2: because of technology I'm sure is not perfect. Some people

3:28 Speaker 2: are going to get caught in that net who are

3:30 Speaker 2: not in fact, uh, you know these AI artists like

3:34 Speaker 2: what was her name, Zani m One we talked about

3:36 Speaker 2: last week or signed So you can't take that or

3:40 Speaker 2: that or that band. Uh.

3:42 Speaker 1: I forget the name of the band now, but we

3:43 Speaker 1: played something.

3:44 Speaker 3: Of theirs Interfury.

3:46 Speaker 2: No Interfury is the one that you found or that

3:48 Speaker 2: guy found you. But there was another band we talked

3:50 Speaker 2: about on the show who uh is charting?

3:53 Speaker 1: Who?

3:53 Speaker 2: People thought? Well people people figured out they're not a

3:56 Speaker 2: real band. But anyway, I forget who the who it

3:59 Speaker 2: is now, but all their songs are like about war,

4:02 Speaker 2: like very pro war but very strange.

4:05 Speaker 1: But yeah, so these are problems.

4:07 Speaker 2: But there's going to be some artists who the software

4:10 Speaker 2: or whatever they're doing, however they're applying this because you

4:13 Speaker 2: can't have a human being sitting there doing all this, right,

4:16 Speaker 2: that software is going to detect artists who aren't necessarily

4:19 Speaker 2: AI generated, but maybe have AI elements in their you

4:23 Speaker 2: know AI. I mean you use AI in a recording

4:26 Speaker 2: studio anyway, you can't avoid it, right, But let's look

4:30 Speaker 2: at this, so it says here again, this is from

4:31 Speaker 2: the Hollywood Reporter. Spotify is set to strengthen AI protections

4:35 Speaker 2: for artists and music producers with a series of measures

4:38 Speaker 2: including improved enforcement of impersonation violations, a new spam filtering system,

4:44 Speaker 2: and AI disclosures for music and industry standard credits. The

4:48 Speaker 2: music streaming Giant made the announcement in a for the

4:51 Speaker 2: record post on its website Thursday, noting that it had

4:55 Speaker 2: removed seventy five million spamy tracks began quote the pace

5:02 Speaker 2: of recent advances in generative generative AI technology has felt

5:07 Speaker 2: quick and at times unsettling, especially for creatives. At its best,

5:12 Speaker 2: AI is unlocking incredible new ways for artists to great

5:15 Speaker 2: music and for listeners to discover it. At its worst,

5:18 Speaker 2: AI can be used by bad actors and consent farm

5:22 Speaker 2: and content farms to confuse or deceive listeners, push slop

5:28 Speaker 2: into the ecosystem, and interfere with authentic artists working to

5:32 Speaker 2: build their careers. That kind of harmful AI content degrades

5:36 Speaker 2: the user experience for listeners and often attempts to divert royalties.

5:40 Speaker 1: To bad actors.

5:42 Speaker 2: UH the future again, this is from the statement the

5:45 Speaker 2: future of music industry. Of the music industry is being written,

5:49 Speaker 2: and we believe the aggressive that aggressively protecting against the

5:53 Speaker 2: worst parts of gen AI is essential to enabling its

5:59 Speaker 2: potential for artists and producers. We envision a future where

6:03 Speaker 2: artists and producers are in control of how or if

6:06 Speaker 2: they incorporate AI into their creative processes. As always, we

6:11 Speaker 2: leave those creative decisions to artists themselves, while continuing our

6:14 Speaker 2: work to protect them against spam, impersonation, and deception, and

6:18 Speaker 2: providing listeners with greater transparency about the music they hear unquote.

6:24 Speaker 2: Regarding specifics on the issue of impersonation, Spotify has committed

6:28 Speaker 2: itself to stronger rules and better enforcement.

6:31 Speaker 1: Quote.

6:32 Speaker 2: We've introduced a new impersonation policy that clarifies how we

6:36 Speaker 2: handle claims about AI, voice clones and other forms of

6:40 Speaker 2: unauthorized vocal impersonation, giving artists stronger protections and clearer recourse.

6:46 Speaker 2: Vocal impersonation is only allowed in music on Spotify when

6:50 Speaker 2: the impersonated artist has authorized its usage.

6:55 Speaker 1: Unquote.

6:56 Speaker 2: By the way, that is, there's more to this article,

7:00 Speaker 2: but that's one of the points where you know, I

7:03 Speaker 2: talked about people getting caught in the net again, I'm

7:06 Speaker 2: not I'm not criticizing Spotify for doing this. I think

7:08 Speaker 2: this is probably a really good thing, but it's not

7:11 Speaker 2: going to be perfect. What if you have an artist

7:13 Speaker 2: who just sounds like they just have a voice similar

7:17 Speaker 2: to another artist. You know, a lot of artists sound similar.

7:21 Speaker 3: And but there's I bet there's things computers can pick

7:24 Speaker 3: up that's different about the may sound similar to the

7:26 Speaker 3: human ear.

7:27 Speaker 1: But maybe this is.

7:28 Speaker 2: Maybe maybe maybe, But but what if I just I

7:35 Speaker 2: can I can see there're being mistakes here. It's not

7:37 Speaker 2: going to be perfect. And what if an artist here's

7:40 Speaker 2: someone who they think sounds like them and they make

7:44 Speaker 2: a complaint, they make a claim, and then it turns

7:48 Speaker 2: out that the artist who sounds like them is a

7:50 Speaker 2: real person with a real voice who happens to sound

7:53 Speaker 2: like them. So there's going to be problems with this.

7:57 Speaker 2: I mean, ask anybody who's dealt with YouTube, you know,

8:01 Speaker 2: getting getting all kinds of you know, the stuff YouTube

8:05 Speaker 2: will put you through if you're a content creator. And

8:09 Speaker 2: we won't get in all of that that because that's

8:11 Speaker 2: a whole other conversation, but it's related in the sense that,

8:14 Speaker 2: you know, sometimes people get caught in a net and

8:17 Speaker 2: it's it's it's not a good uh, I not a

8:22 Speaker 2: good thing to have happened to you.

8:24 Speaker 3: So it's it's going to be kind of a difficult

8:27 Speaker 3: thing to navigate.

8:29 Speaker 1: Yes, yes, it says here.

8:32 Speaker 2: Additionally, Spotify said it was ramping up quote investments to

8:35 Speaker 2: protect against another impersonation tactic where uploaders fraudulently deliver music

8:42 Speaker 2: AI generated or otherwise to another artist profile across streaming services. Uh,

8:50 Speaker 2: testing new prevention tactics with leading artists distributors to equip

8:55 Speaker 2: them to better stop these attacks. As at the source unquote. Yeah,

9:00 Speaker 2: so if you if you make a song that sounds

9:03 Speaker 2: like Taylor Swift and then you try to upload it

9:05 Speaker 2: as Taylor Swift's.

9:09 Speaker 1: Say, this is a new Taylor Swift.

9:11 Speaker 3: Remember, Yeah, people do slimy stuff like that.

9:14 Speaker 2: Oh remember we talked about it on the show Selene

9:17 Speaker 2: Deon where there was this AI generated Celene Dion song

9:21 Speaker 2: that fooled a lot of people and it was being

9:24 Speaker 2: Uh it was pretty successful and I think it even

9:27 Speaker 2: charted on some charts somewhere.

9:28 Speaker 1: I don't know, but it was.

9:29 Speaker 2: It was on YouTube and people thought it was really

9:32 Speaker 2: her and it wasn't horrible. U. Yeah, but uh, but

9:38 Speaker 2: but there's also so sometimes if artists have similar names,

9:43 Speaker 2: I don't remember who it was. It was somebody we

9:45 Speaker 2: had on the show who they They had tried to

9:51 Speaker 2: upload their music to Spotify, but their name was somewhat

9:55 Speaker 2: similar to someone else's name. It wasn't the same name,

9:57 Speaker 2: but it was somewhat similar, and it and it kept it. Yeah,

10:01 Speaker 2: it was causing all kinds of problems. I don't remember

10:03 Speaker 2: the details, but it also says here again. This is

10:08 Speaker 2: from the Hollywood Reporter. Spotify hopes its new spam filtering

10:12 Speaker 2: measures will cut down on issues such as mass uploads, duplicates,

10:17 Speaker 2: SEO hacks, artificially artificially short track abuse, and other forms

10:26 Speaker 2: of slop that have all become easier and more prevalent

10:31 Speaker 2: due to AI tools. The new spam filter quote will

10:35 Speaker 2: identify uploaders and tracks engaging in these tactics, tag them,

10:39 Speaker 2: and stop recommending them unquote. The company says it will

10:43 Speaker 2: roll out a new music spam filter over the coming

10:45 Speaker 2: months and will be careful not to penalize the wrong uploaders.

10:50 Speaker 2: The third measure, Oh yeah, yeah, it won't be perfect.

10:55 Speaker 2: The third measure Spotify has introduced is AI disclosure for

11:00 Speaker 2: music with industry standard credits.

11:03 Speaker 1: This is going to be messy with.

11:05 Speaker 2: AI increasingly being used in the music industry. The company

11:08 Speaker 2: wants to increase transparency of its use. Quote, we know

11:12 Speaker 2: the use of AI tools is increasingly a spectrum, not

11:15 Speaker 2: a binary, where artists and producers may choose to use

11:19 Speaker 2: AI to help with some parts of their productions and

11:22 Speaker 2: not others. The industry needs a nuanced approach to AI transparency,

11:28 Speaker 2: not forced to classify every song as either is AI

11:33 Speaker 2: or not AI. That's that's going to get very, very messy.

11:39 Speaker 2: That's going to be I want to know. That's going

11:42 Speaker 2: to be hard though.

11:43 Speaker 3: But when you have breck now we have record label

11:46 Speaker 3: signing AI artists, well, yeah, how does that work?

11:49 Speaker 1: Well, that's but that's not messy. That's obvious.

11:51 Speaker 2: So it's easy to be transparent about that. But if

11:54 Speaker 2: you're especially if you're creating any type of electronic music

11:57 Speaker 2: in the studio or or at home on your on

11:59 Speaker 2: your laptop, you know, and then how is this that's

12:04 Speaker 2: going to be tricky.

12:06 Speaker 1: I don't know how this is going to work.

12:08 Speaker 2: Spotify says it will help develop and support new industry

12:10 Speaker 2: standard for AI disclosures in music credits that are being

12:14 Speaker 2: developed through the Digital Data Exchange, the International Standard Setting

12:18 Speaker 2: Organization the AI disclosure information will be displayed.

12:21 Speaker 1: Across the Spotify app.

12:25 Speaker 2: Spotify's new AI crackdown comes despite the company embracing AI

12:30 Speaker 2: and other aspects of its business. In February, Spotify said

12:33 Speaker 2: it will accept more AI narrated audio books on its

12:37 Speaker 2: platform through a partnership with eleven Labs. Still, the new

12:42 Speaker 2: AI measures will be welcomed by the major labels and

12:45 Speaker 2: fans after a number of recent news reports of undeclared

12:48 Speaker 2: AI artists racking up thousands of streams on Spotify. In July,

12:53 Speaker 2: The Guardian, Oh, this is the band I was trying

12:55 Speaker 2: to think of. I couldn't remember the name. In July,

12:57 Speaker 2: The Guardian reported the band The Velvet sun Down released

13:01 Speaker 2: two albums and accrued over one million streams on Spotify

13:05 Speaker 2: before it was revealed that the band and its music

13:08 Speaker 2: were AI generated. A Universal Music Group spokesperson told the

13:14 Speaker 2: harl Of Hollywood reporter quote, we welcome Spotify's new AI

13:18 Speaker 2: protections as important steps toward toward consistent forward. I'm sorry,

13:24 Speaker 2: steps forward consistent with our longstanding artist centric principles. We

13:29 Speaker 2: believe AI presents enormous opportunities for both artists and fans,

13:33 Speaker 2: which is why platforms, distributors, and aggregators must adopt measures

13:38 Speaker 2: to protect the health of the music ecosystem in order

13:41 Speaker 2: for these opportunities to flourish. These measures include content filtering,

13:47 Speaker 2: checks for infringement across streaming and social platforms, penalty systems

13:51 Speaker 2: for repeat infringers, chain of custodies, custody certification, and name

13:56 Speaker 2: and likeness verification. The adoption of these measures would inable

14:00 Speaker 2: artists to reach more fans, have more economic creative opportunities,

14:04 Speaker 2: and dramatically diminish the sea of noise and irrelevant content

14:09 Speaker 2: that threatens to drown out artists' voices unquote. All right,

14:13 Speaker 2: So that is that's what's going on at Spotify. And

14:18 Speaker 2: again Spotify is nobody's favorite company again because of the

14:23 Speaker 2: way they pay or don't pay artists. But but I

14:27 Speaker 2: think this is a good thing that they're doing.

14:29 Speaker 1: But we'll see there's going to be some pitfalls with it.