A U.S. appeals court has ruled that Meta Platforms (NASDAQ: META), Alphabet’s Google (NASDAQ: GOOGL, GOOG), ByteDance’s TikTok, and Snap Inc.’s Snapchat (NYSE: SNAP) must face more than 3,000 federal lawsuits alleging their platforms are addictive and target young users. The 9th U.S. Circuit Court of Appeals, based in San Francisco, dismissed an appeal filed by Meta and TikTok on Monday, rejecting the companies' argument that they were immune from such litigation under Section 230 of the Communications Decency Act of 1996.

Legal Ruling Details

The lower court had previously required the social media giants to proceed with the consolidated cases. In its decision, the 9th Circuit clarified that Section 230 serves as a defense against liability for user-generated content, not as blanket immunity from lawsuits alleging harm caused by platform design. Consequently, the appellate court determined that Meta and TikTok had appealed prematurely.

Attorneys Lexi Hazam and Previn Warren, who represent thousands of school districts and individuals in the litigation, stated that the ruling clears the way for state trials to proceed. A separate trial involving claims brought by school districts is scheduled to begin in February.

Broader Legal Landscape

This ruling adds to mounting legal pressures on major technology firms regarding their impact on youth mental health. In a recent development, Meta was ordered to pay $567 million and implement significant changes to its platforms for minors in New Mexico, after a judge ruled the company created a "public nuisance."

Additionally, YouTube, owned by Alphabet, reached a confidential settlement in June with a 16-year-old Florida teenager. The teen alleged that the platform contributed to his social media addiction, sleep problems, anxiety, and depression after he began using it around age 8. The terms of that settlement were not disclosed.

Key Legal Developments

Market Implications

Beyond legal liabilities, these rulings coincide with shifting user behavior that could impact advertising revenue models. A survey by privacy company Incogni revealed that 55% of respondents post less on social media than they did five years ago. This decline in content creation poses a potential challenge for Meta’s advertising business, which relies heavily on a constant stream of fresh user-generated content to maintain engagement.

Meta, Google, TikTok, and Snap did not immediately respond to requests for comment regarding the latest court decision.

DigitalOcean Holdings Inc (NYSE: DOCN) CEO Paddy Srinivasan asserts that open-weight artificial intelligence models are increasingly capable of handling the majority of enterprise tasks, a shift that aligns with Meta Platforms Inc (NASDAQ: META) CEO Mark Zuckerberg’s vision for broad, affordable AI access. Srinivasan told Benzinga in an exclusive interview that these models are balancing the "intelligence per dollar" equation, enabling wider adoption without requiring the highest-cost infrastructure for every application.

The development supports Meta’s recent announcement to offer free versions of its AI tools to billions of users while allowing paid upgrades for additional computing power. Zuckerberg argued in a Monday essay that powerful AI should not be controlled by a small group of institutions. Meta plans to resume releasing some open-source models and has already launched the smaller open-weight Muse Glimmer model for personal computers.

The Economics of Open-Weight Models

Srinivasan highlighted that open-weight models allow businesses to customize systems using their own data and run them on proprietary infrastructure. This approach helps companies retain intellectual property within the enterprise while adapting models to specific real-world usage patterns.

DigitalOcean is building its inference infrastructure around this flexibility, offering access to multiple models from providers including OpenAI, Anthropic, and various open-weight developers. Customers can select models based on cost and performance requirements rather than defaulting to the most expensive options.

Shifting AI Spending Patterns

According to Srinivasan, frontier models — the most capable and typically most expensive systems — are needed for only about 25% of the work among DigitalOcean’s customers. The remaining 75% can often be handled effectively by open-weight alternatives.

This 25/75 split suggests that AI adoption no longer depends entirely on the economics of the most expensive systems. Companies can now choose the "right model, right cost, for every task," avoiding unnecessary expenditure on high-performance computing for routine operations.

What the Numbers Show

The data indicates a structural shift in AI consumption where volume is driven by efficiency rather than raw capability alone. While OpenAI and Anthropic retain relevance for complex reasoning tasks, the majority of enterprise AI utility is moving toward customizable, lower-cost open-weight solutions. This divergence allows organizations to optimize spending by reserving expensive frontier compute for only the most demanding applications, potentially accelerating overall AI integration across sectors that were previously constrained by cost barriers.