Meta Deploys AI Visual Analysis to Detect Underage Accounts

Meta announced Tuesday that artificial intelligence will enforce age restrictions across its social media platforms, targeting users under 13 who should not be accessing Facebook or Instagram at all.

The company's official blog post details the rollout of new age assurance technology designed to detect accounts belonging to minors, even when those users list adult birthdays. This represents a significant escalation from previous self-reporting methods that relied entirely on users providing accurate birth dates during account creation.

According to the Meta blog announcement, the new system combines multiple detection techniques. AI will analyze entire profiles for contextual clues like birthday celebrations or mentions of school grades. The technology scans posts, comments, bios, and captions across various formats including Instagram Reels, Instagram Live, and Facebook Groups.

Visual analysis represents the most technically complex addition. The AI scans photos and videos for visual clues about a person's age that text might miss. Meta explicitly states this is not facial recognition. The system looks at general themes and visual cues—height or bone structure—to estimate someone's general age without identifying the specific person in the image. This distinction matters for privacy compliance, though it raises questions about accuracy thresholds.

When the system determines an account may belong to someone underage, the account gets deactivated. The account holder must then provide proof of age through Meta's age verification process to prevent permanent deletion. This creates a friction point users cannot bypass without documentation.

Reporting flows are being simplified to make it easier for the community to flag underage accounts. Meta is supplementing human review teams with AI models that apply consistent evaluation criteria to every report. In testing, this AI-driven review delivers higher accuracy and faster resolutions than human review alone. The physical experience for users means fewer false positives and quicker account restoration when legitimate age disputes arise.

Geographic rollout follows a staggered approach. While many AI improvements are available worldwide, certain advanced features like visual analysis are currently available in select countries. Teen Account protections are expanding to 27 countries in the EU and Brazil on Instagram. Facebook in the US receives the technology for the first time, followed by the UK and EU in June. Global Instagram expansion is planned throughout the year.

Parents receive notifications in the US on Facebook and Instagram with information about checking and confirming their teens' ages on the apps. The notifications include tips on having constructive conversations with teens about providing correct age information online. Parents globally can access these tools through the Family Center.

This announcement arrives amid significant regulatory pressure. New Mexico Democratic Attorney General Raúl Torrez criticized Meta last week for threatening to shut down social media platforms in the state. A jury in March slapped Meta with $375 million in civil penalties, arguing the website failed to protect minors from sexual predators.

Meta argued that the state's requests for relief are burdensome, broad and "in many cases technologically impractical or completely impossible." The new AI enforcement measures appear designed to demonstrate technical feasibility while avoiding the most invasive verification requirements regulators have demanded.

The $375 million penalty from New Mexico represents a concrete financial consequence for inadequate age verification. Previous enforcement relied on honor systems where users self-reported birth dates. The new approach acknowledges that teenagers frequently misrepresent their age to access features or content restricted to adults.

Teen Accounts have been enrolled for hundreds of millions of teens on Instagram, Facebook, and Messenger since 2024. These accounts include built-in protections that limit who can contact teens and the content they see. Content policies automatically place teens under 18 into a 13+ content setting. The new technology aims to proactively find accounts suspected to be teens even when they list adult birthdays.

Existing age assurance measures include estimating age based on activity and reviewing user reports. If someone attempts to change their birthday from under 18 to over 18, Meta requires verification using an ID or Yoti's facial age estimation tools. The new visual analysis adds another layer to this verification stack.

Technical implementation creates practical challenges. Visual analysis must balance detection accuracy against false positives that could lock out legitimate adult users. The system's reliance on bone structure and height estimation introduces variables like camera angles, photo quality, and image editing that could affect results. Users uploading heavily filtered or edited photos may trigger different age estimates than their actual age.

Privacy advocates will scrutinize the visual analysis component. While Meta states the technology does not identify specific individuals, scanning photos for age-related visual cues still involves processing biometric data. The distinction between age estimation and facial recognition may not satisfy all regulatory frameworks, particularly in jurisdictions with strict biometric data laws.

Industry-wide age verification remains complex. Meta's approach represents one of the more aggressive implementations, but competitors face similar challenges. The technology investment required to maintain these systems is substantial, and ongoing refinement will be necessary as users develop new methods to circumvent detection.

Whether this actually reduces underage access or simply creates more friction for legitimate users remains to be seen. The real test comes when enforcement scales globally and regulators evaluate whether the technology meets their safety standards. Meta's argument about technological impracticality may gain more credibility if these AI systems prove insufficient against determined circumvention attempts.

For now, the announcement signals Meta's willingness to invest in automated enforcement rather than rely solely on user honesty. The question is whether AI can reliably distinguish between a 12-year-old and a 19-year-old when both have uploaded similar content and photos. The answer will determine whether this represents genuine progress or just another layer of compliance theater.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn