America has lost more than 200,000 jobs in creative industries over the last four years, and about 50,000 of those went in the past year alone. That’s according to an analysis from the Apricitas Economics newsletter that’s drawing a lot of attention on Hacker News. Media employment has only fallen this hard, and for this long, during the 2001 and 2008 recessions. There’s no general recession right now, though. What there is, for the first time, is AI that can write novels, make photorealistic images and produce soundalike music at almost no cost.
So is AI killing creative jobs? The honest answer is that it’s hard to prove. The pattern in the data, though, is hard to wave away.
📉 Where the jobs went
The losses are spread out, but some sectors got hit much harder than others:
- Movie and sound recording: down more than 100,000 jobs, nearly a third of the sector. This summer, employment fell to its lowest level since the 2008 recession and came close to a 30-year low. This one sector accounts for almost half of all media job losses.
- Written publishing: down more than 70,000. Part of this is a long decline, with technology grinding away newspaper and magazine jobs for decades.
- Live broadcasting, streaming and graphic design: these are the surprises. They were stable for years and are now shrinking.
The losses are also speeding up, not leveling off. They’ve held at around 50,000 a year and have accelerated in recent months.
🔍 Why AI’s role is hard to pin down
The analysis is careful not to blame everything on AI, and there are a few reasons the picture is murky.
First, media companies keep quiet about how they use AI. Part of that is fear of public backlash. The bigger reason is money: if it came out that AI made their content, the legal basis for their copyrights could weaken.
Second, other forces are hitting the industry at the same time:
- Consolidation in Hollywood
- Content production moving overseas
- Audiences moving from traditional media to social media creators
The most telling clue is a split in the data. Every digital arts subsector is losing jobs, while in-person entertainment keeps growing at a healthy pace. If the problem were weak consumer demand, live events would be shrinking too. They aren’t. That points to something specific to digital content, and cheap generative AI fits that description.
🎸 The Napster playbook
To guess where this goes, the analysis looks back at the first digital shock. Around 2000, Metallica sued Napster after unreleased demos and ripped albums spread across the file-sharing service. Metallica and the RIAA won in court and pushed Napster into bankruptcy. Then they lost the bigger fight anyway.
Piracy made individual songs so cheap that Spotify and Apple Music could buy up huge catalogues and charge a small monthly fee for nearly all recorded music. Streaming revenue still hasn’t caught up to CD-era sales, even before you adjust for inflation. Musicians changed their business model to survive. Recorded music turned into a loss leader that sells concert tickets and merch. Tours got longer, venues got bigger, ticket prices shot up, and artists started acting more like influencers.
What stands out is how neatly that lines up with today’s data. Digital output is turning into a commodity, and the money is moving to experiences you can’t copy.
🧭 What this means for businesses and practitioners
If the internet era is any guide, AI won’t kill creative work. It’ll move where the value sits. Some practical takeaways:
- Media companies: don’t count on copyright lawsuits alone to protect your margins. The RIAA won its case and still lost the market. Put money into things AI can’t easily copy: live events, community and recognizable talent.
- Creative professionals: the work that holds its value is tied to a person, a place or a relationship. Pure digital output is the part that’s getting cheaper.
- AI companies: the copyright and disclosure fights are just getting started. Being clear about where your training data comes from is starting to matter for business, not just reputation.
- Analysts: watch the gap between digital and in-person jobs. It’s the cleanest signal we have for AI’s real effect.
What comes next
The 2000s squeeze on the music industry played out over about 15 years, and it rewarded whoever controlled distribution and live experiences. Generative AI is moving faster, and this time it’s hitting film, publishing, design and broadcasting all at once. Creative businesses that change their model now, instead of fighting it out in court, are the ones likely to come through. The full data breakdown is in the original analysis.