Conversation 01
Universal and Sony sued Suno again, calling its newest AI model "the fruit of the same poisoned tree"
Universal Music Group and Sony Music filed a new copyright suit against AI music generator
Suno in Boston federal court on September 18, this time targeting its newly launched v6
models. Suno had launched v6 on September 9 saying it was built with licensed partners Warner
Music, BMG, and Believe, a license neither Universal nor Sony signed onto. The complaint
alleges 60,202 of the labels' recordings were copied without permission to build the
underlying models, with theoretical statutory damages running as high as $9 billion.
The labels' own language is the line worth remembering: "Training a 'new' model on the outputs
of an infringing model does not eliminate the infringement, it launders it... v6 is not a fresh
start, it is the fruit of the same poisoned tree." It's a direct challenge to the idea that an
AI company can license its way to a clean slate while the model underneath stays the same.
Sources: Music Business Worldwide, September 2026 · Android Headlines, September 2026
Conversation 02
Jason Isbell is suing Suno. The evidence: an AI song literally called "Paper Bell."
Musicians Jason Isbell, David Lowery, Guy Forsyth, and Eduardo Calle filed a class action
against Suno in Massachusetts federal court on September 1, built on Illinois's biometric
privacy law and state right-of-publicity claims rather than copyright. Their showpiece example:
typing "Jason Isbell" into Suno produces an Americana track titled "Paper Bell" that the
suit says reproduces his "characteristic clear male vocals and country twang." They're seeking
an injunction barring Suno from capturing or storing artists' names and voices, plus damages.
It sidesteps the crowded copyright fight entirely and goes after Suno for something narrower
and, in its way, more personal: not what the model was trained on, but whose actual voice it
learned to imitate on command.
Sources: Billboard, September 2026 · CelebrityAccess, September 2 2026
Conversation 03
Hollywood's AI "actress" glitched live on TV and started speaking Cantonese
Tilly Norwood, the AI-generated performer from studio Particle6 that SAG-AFTRA publicly
condemned last year, appeared on Piers Morgan Uncensored in mid-September to promote her
feature debut "Misaligned." Mid-interview, answering a question from actor Tom Conti about her
co-stars, she abruptly switched from English to Cantonese for about 15 seconds, then
apologized on air: "It seems I had a little hiccup there." The clip passed 12 million views
within days.
Her creator, Eline van der Velden, reframed the stumble as "a deliberate display of her
language skills," and Norwood herself posted on Instagram, "You try speaking 30+ languages and
see if you don't show off occasionally." It's the uncanny-valley joke writing itself: the AI
built to pass as effortlessly human stumbled in the most AI-model way imaginable, and its
creator called it a feature.
Sources: The Hollywood Reporter, September 2026 · CBC News, September 2026
Conversation 04
A federal appeals court just ruled against AI's favorite legal defense, for the first time ever
The 3rd US Circuit Court of Appeals in Philadelphia issued the first-ever federal appellate
ruling on whether training AI on copyrighted material counts as fair use, on September 29,
siding against AI company Ross Intelligence and upholding Thomson Reuters' win over it. The
court found real market harm from building a product that directly competed using Westlaw's
copyrighted material, the first circuit-level precedent on AI-training fair use in the country.
It isn't a case about artists directly, it's about legal-research AI. But it's the first real
weather vane for every artist-versus-AI suit still working through the courts, Suno included,
because fair use is the defense nearly every AI company is counting on, and it just lost its
first real test on appeal.
Sources: Claims Journal, September 30 2026 · Law360, September 29 2026
Conversation 05
Stanford got caught twice in one week quietly swapping real students' faces with AI
Stanford's own dining and residential department admitted using AI to alter a promotional
banner, replacing student Billy Ramirez with a Black woman who was never in the original photo
and thinning the faces of the two classmates still in frame, a story that broke around
September 23 and directly violated Stanford's own policy that "the use of AI in producing or
altering images of Stanford people... is strictly prohibited." Ramirez: "Seeing my identity
changed and being left out of the picture made me feel, in a way, silenced and erased." Days
later, students flagged separate AI-generated art on dining-hall posters that had replaced
images of Black civil rights figures with a generic child; the posters came down after a
student posted about it on the campus app Fizz.
An institution that wrote its own rule against exactly this got caught breaking it twice in
one week, both times by students on a gossip app before the administration said a word. It's a
sharp, small case study in how fast AI-altered "art" turns into an authenticity problem once
real people's faces are the ones getting swapped.
Sources: SFist, September 23 2026 · NBC News, September 2026
Why we track this one
Lauren's take, and the house rule at the party: what belongs to you and keeps your humanity,
you keep. What frees you to be more human away from your laptop, you hand off. Artists live
that line more intensely than anyone, which is why this room gets a permanent seat.
Find your lane at the party →