Brazil rewrites internet rules, Apple’s transparency gap and AI moderation workarounds

Hello and welcome to Everything in Moderation's Week in Review, your need-to-know news and analysis about platform policy, content moderation and internet regulation. It's written by me, Ben Whitelaw and supported by paid members like you.

As this lands in your inbox, I'll be landing on the tarmac in London after a few weeks away. I arrive to find a regulator claiming a somewhat hollow victory (see: Policies) and the imminent deadline of the social media ban consultation (EiM #328 and others). What a time to be heading home.

This week's edition isn't all about the UK, or indeed the US, though; among 20+ must-reads and a trio of updates from the T&S community, there are important reads from Brazil and the Middle East. That's what EiM is all about after all.

Welcome to new subscribers from Automattic, Apple, Pioneers Post, Christchurch Call, Electronic Arts, Google, Hinge, University College London and elsewhere around dark — but often still magical — forest that is today's internet.

If writing today's briefing while on holiday isn't enough inspiration to become an EiM member, you also get unadulterated access to EiM's archive of 450+ editions including this week's thought-provoking edition of T&S Insider on standardising parental controls.

Bon voyage for now. Here's your Week in Review — BW

In partnership with the Tech Coalition: Collaboration is driving industry action on child safety

No single company can tackle online child sexual abuse alone. Through the Tech Coalition, 60+ companies are working together to share expertise, develop tools, and strengthen safeguards across the ecosystem.

This collaboration is accelerating progress and raising the baseline for how the industry prevents, detects, and responds to harm. Expanding this collective effort is critical to staying ahead of evolving threats.

JOIN US

Policies

New and emerging internet policy and online speech regulation

Marco Civil Da Internet — Brazil’s lauded civil rights framework for the internet — got its biggest overhaul for a decade this week as President Lula signed two decrees that increase big platforms liability over the content they carry.

The decrees align with a Supreme Court judgment from last year which ruled that platforms must immediately remove hate speech and content that promotes serious crimes. It also gave the National Data Protection Authority (ANPD) — which one tech policy expert told me already has its fair share of critics — authority to oversee the new obligations.

Brazilian professor and tech lawyer Ronaldo Lemos called the justification of the decrees as “the hallmark of the so-called "abusive constitutionalism", which rewrites not only laws but the constitutional rules themselves.” His snapshot analysis is worth reading in full while CNN Brasil has a good guide (in Portuguese).

One from late last week, after Week in Review hit subscribers’ inboxes: Ofcom has welcomed X/Twitter’s commitment to protect UK users from illegal hate and terror content. The commitments were called “a step forward” in a statement but it’s hard not to read the announcement and think: shouldn’t a platform of Twitter/X’s scale already have these safeguards in place? The commitments feel closer to baseline hygiene than a meaningful leap forward — as I’m sure the plethora of non-profits involved in the process can attest to. As such, I question whether this is really the regulatory win it’s being presented as.

Also in this section...

• Board to Examine AI impersonation of British Politician (Oversight Board)
• How countries are regulating AI companion chatbots to protect children (IAPP)
• Cybersecurity Will Swallow Digital Policy in the AI Age (Tech Policy Press)

Parental controls should be standardised

Why make something so important so hard to use?

Everything in Moderation*Alice Hunsberger

Products

Features, functionality and technology shaping online speech

We often hear about users falling foul of AI moderation systems and the false positives they create. So I enjoyed this Wired piece on how Arabic speakers are outsmarting automated moderation by altering spellings, mixing dialects and using coded language to avoid takedowns and shadow bans. Plus I found out what Arabizi means :)

Also in this section...

• GGWP expands its content moderation AI tools beyond games (GamesBeat)
• We Need A More Serious Discussion About Suicide And AI Chatbots (Techdirt)
• How Deepfakes Tore a High School Apart (404 Media)

Platforms

Social networks and the application of content guidelines

Apple did what it often does ahead of its big developer conference and announced a bunch of big 2025 safety numbers, not least that it blocked 2 million problematic app submissions and prevented 1.1 billion fraudulent customer account being created. The other eye-catching number was that it prevented more than $2.2bn in fraudulent App Store transactions across the twelve months, which — presuming we believe their workings out — stands in sharp contrast to other platforms failure to detect and deal with fraud.

Transparency expert, apple-y within: I'm once again left disappointed by Apple's approach here. The news release — and the data put out with it — only goes so far in providing meaningful transparency for a platform with a whopping 850 million weekly (not monthly) users worldwide. The company did publish a dedicated App Store Transparency Report for three years (the last one landing in 2024) but those disclosures now appear to have stalled. If you’re an Apple safety employee reading this and can shed light on why, hit reply and let me know.  

Also in this section...

• Making it easier to understand how content was created and edited (Google)
• Trust, Play, and Platforms: Sharing Lessons for Safer Digital Spaces (NYU Stern)
• Study Shows Why Online Hate Crackdowns Can Backfire (George Washington University) 

People

Those impacting the future of online safety and moderation

Bruce Reed, the new CEO of Common Sense Media’s Youth AI Safety Institute, is stepping into one of the fastest-moving areas of online safety. And he know sit.

Reed has a strong background in politics, having been Joe Biden's White House Deputy Chief of Staff, tasked with co-ordinating the US government’s AI strategy as the technology emerged a few years back. Prior to that, he spent five years as Common Sense Media’s senior tech-policy adviser.

In his new role, he’ll lead an organisation developing standards and testing frameworks for AI systems affecting young people — many of whom have funded the institute. And, as a conversation with Tech Policy Press this week shows, he’s very aware of the political opportunity. Reed called the protecting kids from AI harm  “the most popular, most bipartisan issue in America at this time”.

That’s not a particularly difficult thing in Washington right now but it still underlines how youth safety looks to be the entry point for wider AI regulation.

Posts of note

Handpicked posts that caught my eye this week

• “This high-impact executive role involves driving the long-term T&S product strategy and roadmap, with a specific focus on developing innovative features that set the Hinge brand apart” - Jeff Dunn is hiring for a chunky VP role at Hinge.
• “A different format this time that really helped shape a fantastic day of shared experience, insight and great conversations.” - Online Responsibility Network’s Leanne Proctor reflects on her trip across the Irish Sea.
• "While this decision says nothing about your value, I know that being told you are excellent does not pay rent or quiet the noise at 2am. So here is what I would do in the next 30 days if I were you” - Ahmed Said has some helpful job hunting advice.