Breaking news

Meta’s $18 Billion Child-Safety Deal Puts Age Verification Under Pressure

by News Editor

Meta has agreed to pay up to $18 billion to settle claims from U.S. states over children’s safety on Instagram and Facebook, but implementing the agreement will depend partly on age-verification technology that still faces accuracy and privacy challenges.

The settlement involves 52 attorneys general and requires Meta to introduce major changes to how minors use its platforms. Although Meta has not admitted wrongdoing, most of the measures will remain in place for 10 years, pending judicial approval.

Settlement Goes Beyond The Financial Penalty

The $18 billion payment will be spread over a decade, reducing its immediate impact on Meta, which reported more than $200 billion in revenue in 2025. More significant for the company may be the operational changes required under the agreement.

Teen users will face a default two-hour daily limit across Facebook and Instagram, which can only be disabled with parental permission. Access will also be blocked between midnight and 6 a.m., while notifications will be muted during school hours and users will receive prompts after every 15 minutes of continuous use.

Age Verification Is Central To The Deal

Those safeguards depend on Meta identifying which users are minors. Under the agreement, the company will strengthen its technology for detecting users under 13 and identifying teenagers who may have registered with an adult birthday.

Meta already uses AI-based systems and other signals to identify potentially underage users. The company says it is expanding those systems and will invest in stronger age-assurance technology under the settlement.

Accuracy remains a challenge, however. Age-assurance systems can mistakenly classify adults as minors or fail to identify children, while different verification methods create different privacy risks.

Privacy Creates A Second Challenge

Current approaches can include government ID checks, facial age estimation and other forms of identity or behavioural analysis. Each method requires companies to balance accurate age checks against the amount of sensitive information users must provide.

A breach involving identity documents or biometric information could create serious consequences, particularly for minors. Unlike a password, biometric information cannot simply be changed after it is compromised.

Some experts argue that companies can reduce those risks by verifying age without retaining the underlying identity information, for example by generating a token that confirms whether a user falls below a particular age threshold.

Other Platforms Face Similar Pressure

Recent attempts to introduce age verification show how difficult implementation can be. Discord delayed its global rollout earlier this year following user backlash and said it would add alternative verification methods before expanding the system further.

Meta is now calling on TikTok and YouTube to adopt similar protections. The settlement gives that push an additional financial incentive: about 30%, or roughly $5.3 billion, of Meta’s payment is contingent on the two platforms introducing specified measures and making matching payments.

Settlement Could Set A New Platform Standard

For Meta, the agreement represents a significant shift in how child safety is built into its platforms, with several protections becoming default rather than optional.

The bigger test will be whether age assurance can identify minors accurately enough to make those safeguards effective without requiring users to surrender excessive personal information. If it succeeds, the settlement could establish a broader standard for how major social platforms handle children’s access and safety.

Google’s Gemini Has A Branding Problem As AI Apps Grow More Complicated

by News Editor

Google’s latest Gemini update highlights a broader problem in consumer AI: companies are increasingly turning internal tools and capabilities into separate products that users must learn to navigate.

In its announcement of new Gemini Live voice features, Google said users should not have to determine whether a task requires Spark, Daily Brief or a simple inbox search. Yet those are precisely the distinctions the Gemini app currently asks users to make.

Too Many Features, Too Many Names

Gemini users can switch between Chat, Spark and Daily Brief, each with its own icon and place in the app. Rather than simplifying the experience, the growing list of branded features risks making the underlying technology more visible than it needs to be.

Daily Brief illustrates the problem. Google describes it as a source of personalised, proactive updates based on information from services such as Gmail and Calendar. In practice, however, some of its suggestions can feel less like useful assistance and more like unsolicited reminders about previous searches or unfinished research.

Spark has almost the opposite problem. The feature can act as an AI agent capable of completing tasks on a user’s behalf, but packaging that capability under a separate brand forces users to understand when and where they should use it.

A simpler approach would be to let users describe what they need and allow Gemini to determine whether a standard response, an agent or another capability is appropriate.

Gemini Is Not Alone

Google’s approach reflects a wider trend across the AI industry, where companies increasingly expose the architecture of their products through separate modes and branded features.

Anthropic, for example, asks users to distinguish between Claude’s standard chat experience and Cowork. ChatGPT similarly separates Chat and Work. For consumers, these distinctions can turn what should be a simple interaction into a question about which product or mode to use.

That approach is largely driven by how AI systems are built, rather than by how people naturally think about using them.

Apple Takes A Different Approach

Apple’s strategy for Siri offers a contrasting model. Rather than requiring users to learn a new AI interface, the company is integrating AI capabilities into tools people already use, including Spotlight, Photos, the camera and voice requests.

That approach could prove more effective as AI becomes a mainstream consumer technology. Users do not necessarily need to understand which model, agent or feature is handling a request; they simply need the system to complete the task.

Text-Based AI Offers A Simpler Model

The popularity of text-based AI assistants points in the same direction. Services such as Poke, Ollie, Lindy, Orchid, Lucas, Folk, Tomo and Instinct largely reduce the interaction to a familiar interface: send a message and let the assistant determine what needs to happen next.

That simplicity removes an additional layer of decision-making. Users do not need to choose between Chat, an agent or a specialised feature before asking for help.

As a16z investment partner Justine Moore recently argued, consumers increasingly want an AI assistant to feel like a contact they can message rather than another application they must learn.

For Google and its competitors, the challenge may therefore be less about adding capabilities and more about hiding the complexity behind them. The AI that wins mainstream adoption may ultimately be the one that asks users to understand the least.

Uol
eCredo
The Future Forbes Realty Global Properties
Aretilaw firm

Become a Speaker

Become a Speaker

Become a Partner

Subscribe for our weekly newsletter