The US Copyright Office has opened a public inquiry into music streaming fraud. Its notice, dated October 5 and published in the Federal Register on October 7, asks how the fraud is carried out, what it costs and how it connects to AI-generated music. It names the main harm as "the dilution of the royalty revenue pool." The inquiry follows a May 21 letter from a member of the House Judiciary Committee.
Written comments are due November 23 and reply comments December 21. The notice cites Deezer's figures, including that up to 85% of streams on fully AI-generated tracks were fraudulent in 2025, and UMG's September 15 lawsuit against DistroKid. It notes that DistroKid, TuneCore, Believe and UnitedMasters had not all signed the Streaming Integrity Initiative when it launched on September 14.
A regulator is asking the industry to measure a problem the industry has not yet agreed how to count.
I learned this building airplay monitoring for thousands of radio and internet stations: detection is never the hard part; a market forms when everyone agrees it is required and one trusted party does the counting.
The open question is whether the industry files a shared evidence standard by November 23, or a set of competing numbers.
A federal judge in Manhattan sentenced the North Carolina man behind an AI-music streaming fraud to 18 months in prison and two years of supervised release on October 6, and ordered forfeiture of $8,091,843.64, the loss figure both sides agreed. He pleaded guilty in March to one count of conspiracy to commit wire fraud.
Prosecutors had asked for at least 46 months, the low end of the 46-to-57-month guideline range. The probation office recommended 24 months and his lawyers asked for probation.
The Operator’s Take: The sentence is under 40% of the 46-month minimum prosecutors sought, so criminal law will deter only some of this on its own. The practical control point stays the platform and distributor rules on uploads and payouts.
Build · Detection & Telemetry Set automatic review triggers for tracks with heavy play from few accounts, and hold payouts on flagged tracks until they are checked.
At Advertising Week New York on October 5, TikTok made its ad network, formerly Pangle, generally available for US campaigns through Ads Manager. It is also testing Buy Direct, a one-click in-app checkout, and a Shopping Assistant agent, with Salesforce, Shopify, Shoplazza and Stripe as partners, plus an AI lead-generation agent.
A TikTok for Business connector now works inside Claude, Perplexity and Manus. No pricing was given, and the performance figures are TikTok's own.
The Operator’s Take: In Buy Direct the checkout, the payment and the assistant sit with different companies, so whoever holds the record of the sale holds the leverage. I learned this running embedded commerce programs across a dozen countries: multi-party revenue works only when every party reads the same transaction record.
Commercial · Economic Model Before joining an in-app checkout test, put in writing who owns the customer record and what order, return and revenue reporting you receive; do not accept platform summaries alone.
At its unBoxed event, Amazon made its DSP a default agentic chatbot experience and renamed the Amazon Ads console the Amazon Ads Agent. Branded Conversations turns sponsored prompts into multi-turn shopping chats built on brand-supplied colors, fonts, imagery and product descriptions, and Conversational Briefs let advertisers inform Amazon's AI recommendations.
Request Reviews ads are priced per click, and advertisers pay even for poor reviews. Brands cannot choose which reviews the AI surfaces. No pricing or availability dates were given for the other tools.
The Operator’s Take: Brands are now briefing the retailer's AI rather than targeting a shopper, and Amazon says it will give feedback to advertisers who break the guidelines or submit false information. Advertisers pay per click for reviews they cannot steer.
Commercial · Deal Execution Ask Amazon for reporting on how often, and in what words, its assistant repeats your Conversational Brief, and keep a dated copy of every brief you submit.
Independent agency InterMedia now runs its agentic workflow on most of its biddable CTV campaigns, using Claude as its dashboard and connecting it by MCP and API to Vibe, a self-serve CTV platform recently acquired by Walmart, and Olyzon, a CTV decisioning start-up. More than 1,000 brands have adopted Vibe's MCP integration.
The agent can change performance settings when directed, but the agency still verifies settings in each DSP first. Olyzon's CEO says APIs give more consistent, repeatable results than MCP.
The Operator’s Take: The agency owns the dashboard and rents the execution, so the settings check inside each DSP is the only audit left. The agency keeps the harness; the platforms keep the data.
Build · Detection & Telemetry Require every agent-made campaign change to write a log entry (what changed, when, before and after) that you can read outside the platform.