The World's Top 10 Tech Policy Stories This Week
Ten stories, eight jurisdictions, one question running under all of them: who gets to consent, and who gets to say no? Here's the week in tech policy from around the world, followed by what it means.
Meta Pays $16.68B to Settle Teen Safety Claims
Meta disclosed on August 26 it will pay up to $16.68 billion to resolve claims from 29 states that it designed Facebook and Instagram to be addictive to minors. It must now roll out daily time caps, overnight restrictions, and stronger age verification nationwide.
Yahoo Finance →EU's AI Act Transparency Rules Kick In
As of August 2, AI systems must disclose when a user is talking to a machine and label deepfakes as artificial. But the Act's toughest rules — covering biometrics, employment, and essential services — were pushed back to December 2027.
Al Jazeera →Nigeria Investigates Meta, Google, and X Over News Content
Nigeria's competition authority opened an investigation into Meta, Alphabet, and X — plus unnamed generative AI platforms — over uncompensated use of copyrighted news content, as its data protection regulator signals it will intensify enforcement.
Concurrences →China Fines 12 Firms in First Week of Companion-AI Rules
China's new framework treating AI agents as a distinct legal category took effect July 15. Within the first week of enforcing its companion-AI rules, regulators fined 12 companies a combined 4.2 million RMB.
IAPP →Brazil Puts ChatGPT, Claude, and TikTok Under Monitoring
Brazil's data authority began formally monitoring 22 platforms and AI tools — including Instagram, TikTok, and major chatbots — for compliance with children's data protections, part of a 2026–27 plan that includes 20 targeted AI-practice inspections.
MLex →South Korea May Let AI Firms Train on Data Without Consent
An amendment cleared committee with bipartisan support that would let companies use personal data to train AI beyond its original collection purpose — without the person's consent — if the privacy regulator approves. A floor vote was expected as early as mid-August.
IAPP →India Moves Toward an AI Liability Framework
Public feedback on the Digital India Act has closed, and reporting indicates a statutory framework establishing legal liability for AI-related harms is being finalized ahead of the fall parliamentary session — a notable step for the world's most populous country.
Responsible AI Labs →70% of Americans Oppose a New AI Data Center Nearby
A Gallup poll found 70% opposition to new AI data centers in respondents' own neighborhoods — more than oppose a nuclear plant nearby. Pennsylvania's governor signed an executive order tightening data center development standards as water use and electricity prices become flashpoints.
CNBC →UK-South Korea Trade Deal Bans Forced Data Localization
Negotiators concluded an upgraded free trade agreement with a dedicated digital trade chapter, guaranteeing cross-border data flows and prohibiting unjustified data localization requirements.
techUK →FTC Takes Zillow and Redfin to Trial
The FTC's case accusing Zillow and Redfin of colluding on rental advertising went to trial on August 21 — one of several fronts testing how far competition law reaches into platform-driven markets, alongside Australia's parallel case against Meta opening September 1.
MLex →What This Means
This week's ten stories span five continents, but they all turn on the same question: who gets to consent, and who gets to say no?
In South Korea, lawmakers are moving to let AI companies train on personal data without asking the person first — as long as the privacy regulator signs off. In Brazil, regulators are watching the same chatbots, and the platforms that host them, specifically because children can't meaningfully consent to how their data gets used. And in the United States, the fight isn't over data at all — it's over land, water, and electricity. Seventy percent of Americans now say they don't want an AI data center built near them, more than oppose a nuclear plant. That's not NIMBYism. That's a public that watched a decade of "trust us" become the default answer to every extraction of value from their communities, and decided this time to ask first.
I'll say plainly where I come down on this, because I think the throughline demands it: data centers are, on net, bad for the communities that host them. They strain water tables, spike electricity bills, and get sited with a fraction of the community input a highway or a power plant would require. Pennsylvania's governor signing an executive order to rein that in isn't an overreaction — it's regulators catching up to a harm that's been externalized onto the public for years. And yet: we need data centers. There is no version of the AI systems now being folded into hiring, healthcare, and government decision-making that doesn't run on that physical infrastructure somewhere. Holding both of those truths at once — that AI's infrastructure is necessary, and that its current build-out is extracting cost from people who never agreed to pay it — is, I think, the only honest place to stand on this issue right now.
That same tension runs through the enforcement stories this week. Nigeria isn't waiting for a global consensus on AI and copyright before opening an investigation into Meta, Alphabet, and X over uncompensated use of news content — its competition authority decided the harm was clear enough to act on now. China fined a dozen companies in the first week of enforcing its new companion-AI rules. Brazil didn't wait for its long-stalled AI bill to pass before putting ChatGPT and Gemini under formal monitoring for how they handle kids' data. Meanwhile, the EU's AI Act — the most comprehensive framework on paper — just pushed its hardest obligations back to December 2027, and the U.S. still has no federal privacy law at all.
The pattern is hard to miss: the jurisdictions actually holding Big Tech accountable in real time are not necessarily the ones with the most sophisticated legislative frameworks. They're the ones willing to act on the authority they already have, imperfect as it is, while the "gold standard" frameworks keep pushing their teeth further into the future. If you care about data rights and genuine accountability rather than the appearance of it, that's the trend worth watching this fall — not which country passes the most elegant law, but which regulators actually show up.
Worth watching next: South Korea's floor vote on AI training-data consent, the FTC v. Zillow/Redfin and ACCC v. Meta trials, and whether Nigeria's news-content investigation produces a real enforcement action or a settlement like its earlier Meta case.