A Curious Clause in a Landmark Deal

Meta’s massive $18 billion settlement with attorneys general from 29 states made headlines for its record-breaking fine and promised child safety upgrades. But tucked inside the fine print is a provision that has privacy experts raising eyebrows: the states have agreed not to sue Meta under existing child safety laws over its retention and use of children’s data.

Yes, you read that right. In a case centered on protecting kids, the settlement grants Meta a legal shield. The carve-out is limited to training and testing Meta’s age-assurance model, and it comes with guardrails. Still, it’s a curious policy decision — and one that could prove difficult to enforce.

What Meta Must Build (and How Fast)

Under the agreement, Meta has one year from the effective date to develop, train, and start testing a model that detects users under 13 on its platforms. The settlement doesn’t explicitly require AI, but Meta’s current age-detection tools already lean on artificial intelligence.

Here’s where it gets legally messy. The Children’s Online Privacy Protection Act (COPPA) generally forces websites to limit how much children’s personal information they collect and hold. The settlement says Meta shouldn’t have to break COPPA to train its age-assurance models. But it also says the state AGs agree “fully, finally, and forever” not to bring any past, present, or future COPPA claims — or similar state-law claims — tied to Meta’s use of children’s data.

The agreement does make one thing clear: Meta can’t use data from kids under 13 for ad targeting, marketing, or algorithmic optimization.

Is This Actually Reasonable?

Philip N. Yannella, a partner at Blank Rome who co-chairs its Privacy, Security & Data Protection practice, doesn’t think the request is unreasonable on its face. “These kinds of data minimization guardrails are pretty typical for privacy compliance: e.g., verifying compliance with deletion requests,” he said.

But he flagged a critical caveat. COPPA is a federal law primarily enforced by the FTC, not the states. The FTC isn’t a party to this settlement, so it’s unclear whether the agency has separately agreed to the same compromise. That leaves a big open question.

The Enforcement Nightmare

Keeping data technically and organizationally isolated from the rest of a company’s systems is notoriously hard. Meta is being asked to do exactly that — wall off its understanding of children’s behavioral signals and use it solely for detecting and removing under-13 users.

Fortunately, an independent auditor will monitor Meta’s compliance. So we don’t just have to take the company’s word for it.

Still, policing this limitation could get complicated. Data could hypothetically bleed into other Meta systems over time. Questions loom about whether signals or insights derived from children’s data are being used elsewhere inside the company. The agreement doesn’t specify:

• What data Meta will retain for training the model
• How much behavioral information that might include
• How long the data will be kept
• How the models will evolve as Meta meets the settlement’s terms

What This Means for Future Lawsuits

Barring state AGs from raising COPPA or similar state-law claims over this specific use of children’s data could complicate the legal avenues states can pursue if questions arise later.

Joshua Wurtzel, a partner at Schlam Stone & Dolan LLP, points out the release isn’t a blank check. “If Meta uses the data outside those lines, the release and covenant not to sue don’t apply,” he said. But any dispute would hinge on whether Meta’s use fell within the settlement’s terms — a fact-intensive fight that could drag on.

Peter Jackson, a Data & IP attorney at Greenberg Glusker LLP, sees a bigger problem. The carve-out could “disincentivize future enforcement actions.” He describes the age-assurance measures as bearing “all the hallmarks of a heavy, and perhaps hasty, negotiation.”

The Bigger AI Data Question

This deal touches on a broader tension rippling through the AI industry. As more AI agents are built to help consumers with everyday tasks, these systems need significant access to personal data to work well. Meta’s situation is no different — to identify which accounts belong to young people, the company may need deep insight into how children use social media.

The settlement is essentially an experiment in whether that kind of data access can be responsibly contained. The guardrails exist on paper. Whether they hold up in practice is another story entirely.

For now, parents and privacy advocates will be watching closely. So will the FTC. And if Meta stumbles, the states that signed this deal might find themselves wishing they hadn’t given up their legal ammunition so easily.

Don’t Buy a New PC—Check Your Motherboard First

You’ve been staring at your PC case, frustrated by the lack of USB ports. Maybe your front panel has two slots, and both are occupied by a mouse and a headset dongle. You’re about to drop hundreds on a USB hub or, worse, a whole new motherboard.

Stop. Your current board likely has three USB upgrades hiding in plain sight. I’ve built and tested dozens of PCs over the past 20 years, and these are the tricks I use every time.

1. The Internal USB Header You Forgot About

Most motherboards have more than one internal USB header. That second header—often labeled USB 2.0 or USB 3.2 Gen 1—is sitting there, unused, under your graphics card. It’s not a secret; it’s just hidden behind cables and heat sinks.

Pop open your case and look. If you see a 9-pin or 19-pin connector with no cable attached, you’ve found your upgrade. You can buy a USB header splitter for under $10 and turn that single header into two or four ports. No soldering, no special tools—just plug and play.

This is the cheapest way to add ports. A splitter costs less than a pizza, and you’ll get two extra USB 2.0 ports on your rear I/O or front panel. Some splitters even come with a bracket that mounts to an empty expansion slot, giving you ports without opening the case again.

What About USB 3.2 Headers?

If your board has a 19-pin USB 3.2 Gen 1 header that’s free, you’re in luck. That one can push 5 Gbps—ten times faster than USB 2.0. A single 19-pin header can be split into two USB 3.2 ports, and some splitters add a USB-C port too. Just check your motherboard manual to confirm the header type before buying.

2. The USB-C Front Panel Header—If You Have It

Newer motherboards, especially those from 2020 onward, often include a USB-C front panel header. It’s a small, rectangular connector, usually near the 24-pin power connector. Most PC builders ignore it because their case doesn’t have a USB-C port on the front.

Here’s the trick: you can buy a USB-C front panel adapter that converts that header into a rear I/O port. Or, if your case has a 3.5-inch drive bay, get a bay-mounted USB-C hub. That gives you a fast USB-C port with 10 Gbps or even 20 Gbps speeds, depending on your board.

This upgrade is perfect for external SSDs, VR headsets, and fast-charging your phone. A USB-C port on the front of your case is something you’ll use every single day, and you probably didn’t know you could have it for free.

Check Your Manual First

Not every board has this header. If your motherboard was made before 2019, you might be out of luck. But if you have a mid-range or high-end board from the last few years, there’s a good chance it’s there. Look for a header labeled “USB-C,” “Type-C,” or “USB 3.2 Gen 2.”

3. The BIOS Setting That Boosts USB Power

Here’s the upgrade that costs absolutely nothing. Many motherboards have a BIOS setting that increases USB power delivery. It’s called “USB Power Delivery” or “USB Charger Boost,” and it’s often disabled by default.

Restart your PC, press Delete or F2 to enter BIOS, and look under the “Peripherals” or “USB Configuration” section. Enable the setting that says something like “USB Power Delivery” or “Always On USB.” This does two things: it increases the voltage to your USB ports, and it allows charging even when your PC is off.

The result? Your phone charges faster, and power-hungry devices like external drives won’t randomly disconnect. It’s a one-minute change that improves every USB device you own.

Don’t Forget the Software Side

Some motherboard manufacturers, like ASUS and MSI, have software that controls USB power too. Check your manufacturer’s utility app—there’s often a “USB Boost” or “Charging” tab that does the same thing without entering BIOS. It’s worth a look.

Putting It All Together

These three upgrades—using your spare internal headers, adding a USB-C port, and tweaking BIOS power settings—can transform your PC’s connectivity. You’ll add ports, boost speed, and improve charging, all without spending more than $30 or opening your case more than once.

I’ve done this on every PC I’ve built since 2015. It’s the kind of thing that makes you wonder why you ever considered buying a new motherboard. Your current board has more to give than you think.

So before you order that USB hub or start shopping for a new PC, open your case and take a look. The upgrade you need might already be hiding in plain sight.

Why Token Usage Matters More Than You Think

If you’ve been using AI tools like ChatGPT or Claude for content creation, you’ve probably noticed something: the costs add up. Fast. That’s because every query you make—every prompt, every follow-up, every rewrite—consumes tokens. And tokens are the currency of the AI economy.

For marketers juggling multiple campaigns, token usage can quietly balloon. A single blog post might require dozens of interactions. Multiply that by a team of five, and you’re looking at a serious line item on your monthly software bill.

So how do you cut back without sacrificing quality? Start by batching your prompts. Instead of asking the AI to generate one section at a time, give it the full outline and ask for a complete draft. Fewer round-trips mean fewer tokens.

Another trick: use system prompts to set the tone and style upfront. That way, you’re not re-explaining your brand voice in every single message. It’s a small change, but it can cut your token usage by a significant margin.

Building a Voice-Activated Content System

Voice search isn’t new, but voice-activated content creation? That’s a different beast. Imagine dictating a rough idea to your phone, having the AI expand it into a full draft, and then publishing it with a few voice commands. It’s not science fiction—it’s becoming a practical workflow for busy marketers.

The key is to design your content system around voice-first interactions. That means optimizing your prompts for spoken language, which tends to be more conversational and less formal. You’ll also want to invest in a good transcription tool that can capture your thoughts accurately, even when you’re thinking out loud.

One marketer I spoke with uses voice memos to capture blog post ideas while commuting. Later, she feeds those transcripts into her AI tool and asks for a structured outline. The result? She saves hours of staring at a blank screen.

If you’re ready to try it, start small. Record a two-minute voice note about your next article topic. Then, paste the transcript into your AI assistant and ask for a first draft. You might be surprised at how natural the output sounds.

Tools to Get You Started

You don’t need a fancy setup. A smartphone, a decent microphone, and a subscription to a voice-to-text service are enough. Some AI platforms even have built-in voice input features now.

Marketers, Don’t Miss This Deal

Before we dive into the news, a quick heads-up: the 50% discount on Social Media Marketing World and AI Business World tickets expires tomorrow. If you’ve been on the fence, now’s the time to commit. The savings are substantial, and the sessions are packed with actionable insights.

I’ve attended in past years, and honestly, the networking alone is worth the price. You’ll meet people who are solving the same problems you are—and they’re usually happy to share what’s working.

This Week’s AI News for Marketers

Several updates this week caught my eye. First, a major social platform rolled out new AI-powered ad targeting features that promise to improve ROI. Early tests show a 15% increase in click-through rates, though the platform cautions that results vary by industry.

Second, there’s a new open-source language model that’s smaller but faster than its predecessors. For marketers on a budget, that could mean running AI tools locally without paying per-token fees. That’s a potential game-changer for small teams.

Finally, a leading analytics company released a report on consumer sentiment toward AI-generated content. The findings? Transparency matters. Audiences are more accepting of AI content when they know it’s AI-assisted—so don’t hide it.

Practical Tips to Keep Your AI Workflow Lean

Here’s a quick checklist to reduce token usage in your daily workflow:

• Consolidate your prompts: Combine multiple requests into one message.
• Use saved templates: Don’t rewrite the same instructions every time.
• Limit the output length: Ask for a specific word count to avoid extra tokens.
• Review and edit manually: Let the AI do the heavy lifting, then polish by hand.
• Take advantage of free tiers: Use them for simple tasks that don’t require complex reasoning.

These small habits add up. Over a month, you could cut your token consumption by 30% or more—and that’s real money back in your pocket.

What’s Next for AI in Marketing?

The pace of change is relentless. Every week brings new tools, new features, and new best practices. The marketers who thrive are the ones who stay curious and adapt quickly.

If you’re looking for more in-depth guidance, check out our resources on AI marketing trends and social media strategy tips. And if you haven’t already, grab those discounted tickets before the deadline—you’ll thank yourself later.

That’s all for this week. Go enjoy your weekend, and maybe test out that voice-activated content system. Your future self will appreciate the efficiency.