Los Angeles Superior Court Judge Carolyn Kuhl denied motions for a new trial on Tuesday, according to court documents.

The ruling means a March jury verdict awarding $6 million in damages will remain in place while both companies pursue appeals.

The case was brought by a 20-year-old California woman identified in court records as K.G.M., also known as Kaley.

She told jurors she began using YouTube at the age of six and Instagram at nine, and later developed anxiety, depression, body dysmorphia and suicidal thoughts.

Her lawyers argued that features built into the platforms, including algorithmic recommendations, beauty filters, endless scrolling and push notifications, encouraged compulsive use and contributed to her mental health issues.

After hearing the evidence, the jury found both companies negligent and concluded they acted with malice, oppression and fraud.

Jurors awarded $3 million in compensatory damages and a further $3 million in punitive damages, bringing the total award to $6 million.

Meta was assigned 70% of the liability, amounting to $4.2 million, while YouTube was held responsible for the remaining 30%, or $1.8 million.

The trial attracted attention because it was the first to reach a verdict among more than 1,600 related lawsuits filed across the United States by young people, families and school districts.

The litigation accuses social media companies of designing products that encourage addiction among children and teenagers while contributing to mental health problems.

Several senior technology executives testified during the proceedings. Meta chief executive Mark Zuckerberg spent about eight hours on the witness stand and was questioned about internal company documents showing that Instagram had four million users under the age of 13 in 2015.

Instagram head Adam Mosseri also testified and acknowledged that spending 16 hours a day on the platform could be “problematic.”

Meta said it “respectfully disagrees” with the verdict and plans to appeal. The company argued that teenage mental health is influenced by many factors and cannot be linked to a single app.

Google, which owns YouTube, also intends to challenge the ruling. The company argued that the case “misunderstands YouTube” because it views the service as a streaming platform rather than a social media network.

As it stands, lawmakers and child safety advocates are currently pushing for stronger protections for young users online, including uncompromising age-verification requirements, expanded parental management and changes to platform design.

The company announced the changes on Monday, making AI Plus the lowest-priced paid AI subscription offered by a provider in the US market.

Vikas Kansal, product lead for Gemini AI subscriptions, said on X that the storage upgrade would reach users over the next few days.

Google AI Plus was introduced in January as an entry-level paid plan aimed at individual users and students. The service includes access to Gemini with higher usage limits, Omni Flash video generation, Google Flow creative tools, NotebookLM and AI-powered features in Gmail.

In Nigeria, alongside AI Plus at N7,700, Google still offers higher-priced plans. Google AI Pro costs N28,500 per month and includes 5TB of storage, expanded Gemini access and the company’s Pro model.

Google AI Ultra starts at N89,000 per month, offers at least 20TB of storage and provides significantly higher usage limits, as well as early access to new features.

The current price reduction follows a series of changes to Google’s AI subscription business this year. In April, the company increased storage on its AI Pro plan to 5TB without raising prices. A month later, it launched a new AI Ultra package and reduced the cost of its top-tier subscription from $250 to $200 per month.

With competition increasing among AI providers over subscription pricing, and premium plans taking over the market, companies have now started introducing cheaper options to attract more users.

This first became visible in India, one of the world’s fastest-growing AI markets. OpenAI launched ChatGPT Go there in August 2025 at about $4.60 per month, well below the price of its standard ChatGPT Plus subscription. Google followed with its own sub-$5 AI Plus offering in India later that year.

Google’s latest decision brings that pricing strategy to the United States, where subscription costs have so far played a smaller role in competition between major AI companies.

The development could increase pressure on competitors, particularly Anthropic, which has not introduced a lower-cost subscription tier or localised pricing in key international markets.

OpenAI and Anthropic are both preparing for public listings after filing confidential IPO paperwork, and growing price competition could become an important issue for investors assessing the long-term profitability of AI businesses.

The company said it will start using internal detection systems from May 2026 to identify videos containing what it described as “significant photorealistic AI” content.

When creators fail to disclose that material, YouTube will now add the label automatically.

The update also changes where viewers see those warnings. Instead of hiding them inside video descriptions, YouTube will place labels directly below long-form videos and over Shorts, making them easier to spot.

YouTube has required creators since 2024 to disclose content made with AI tools when videos could realistically be mistaken for real people, places or events. However, content that was clearly fictional, animated or unrealistic did not need the same treatment.

Now, the company says it wants a more reliable system as AI video tools become harder to distinguish from real footage.

“We’ve heard consistently from our community that they value transparency when it comes to generative AI content,” YouTube said.

“That’s why since 2024, we’ve been labeling content when creators disclose they’ve used AI tools.”

The platform said the policy itself has not changed, but enforcement is becoming more active as AI-generated video quality improves.

The announcement follows the launch of Google’s Gemini Omni models at the company’s developer conference last week. Google said the models can generate highly realistic videos while showing an understanding of subjects including physics, science, history and culture.

Under the new system, creators will still be expected to disclose AI-generated content themselves. However, YouTube explained it will step in when its systems detect realistic AI content that has not been labelled.

“If a creator doesn’t specify whether or not they used AI, but our systems detect significant photorealistic AI use, we will now automatically apply a label,” the company said.

Creators who believe their content was wrongly flagged will be able to update the disclosure status through YouTube Studio. Still, YouTube said labels will remain permanent in some situations.

That includes videos produced using YouTube’s own AI tools such as Veo and Dream Screen. The same applies to videos carrying C2PA metadata showing they were fully generated with AI systems.

C2PA is an industry standard designed to help identify AI-generated and digitally altered media. Companies including OpenAI, Nvidia, Kakao and Eleven Labs have backed the standard in recent months.

YouTube is also changing how labels appear across the platform.

For long-form videos, labels will now be directly below the video player and above the description section. On Shorts, viewers will see them as overlays on the video itself.

The company said labels for unrealistic or lightly edited AI content will still appear only inside the expanded description section.

“By moving these labels on to the main stage, viewers get the context they need at a glance,” YouTube said.

The changes align with YouTube’s expansion of other tools aimed at detecting manipulated content. The company recently increased access to its AI deepfake detection system, allowing adults to scan the platform for videos that may contain their likeness.

At the same time, YouTube continues adding AI features across its services, including AI-generated video summaries, playlist tools for YouTube Music, interactive search functions and creation tools for creators.

Despite the labelling system, YouTube said the presence of an AI label will not affect recommendations or whether creators can make money from their videos.

“It’s important to note that a disclosure label alone does not change how a video is recommended or whether it’s eligible to earn money,” the company said.

People familiar with the talks say investors have already made early offers in the range of $850 billion to $900 billion but the company has not accepted any of them.

Discussions are still at an early stage, and nothing has been agreed.

Several investors are trying to secure large stakes, with some offers said to be worth up to $50 billion in new capital. A decision is expected at a board meeting in May.

Anthropic last raised funds in February and that round brought in $30 billion, valuing the company at $380 billion. If this new round goes through at the higher valuation being discussed, it would mark a surge in a matter of months.

The move would also change its position in the market as OpenAI was valued at about $852 billion in March after a major funding round, but a deal at $900 billion would place Anthropic ahead as the most valuable artificial intelligence startup.

The company has received backing from major technology firms. Google has committed billions of dollars, with more funding tied to performance targets. Amazon has also invested heavily and plans to increase its stake over time.

Anthropic declined to comment when contacted.

Revenue growth has supported the surge in investor interest with the company’s annual revenue run rate passing $30 billion earlier this year and is now said to be approaching $40 billion.

Growth has come from demand for its Claude models, especially tools built for coding and business use.

Recent releases include new versions of its core systems and a cybersecurity-focused model with limited access due to safety issues.

There is also a public listing under consideration. Bloomberg reported that an initial public offering could come as soon as October. If that plan holds, this funding round may be the last before the company goes public.

The agreement places Google alongside OpenAI and Elon Musk’s xAI as companies now supplying AI tools for sensitive military use.

Under the deal, the Pentagon can use Google’s AI for “any lawful government purpose”. That can include work carried out on classified networks, such as mission planning and weapons targeting.

The report said Google must also help adjust some of its AI safety settings and filters if requested by the government.

At the same time, the contract includes limits on how the technology should be used. It states that the AI system is not intended for domestic mass surveillance or autonomous weapons, including target selection, without proper human oversight and control.

However, the agreement reportedly also says Google cannot block or overrule lawful operational decisions made by the government.

Google said it continues to support public sector customers across both classified and non-classified environments.

A company spokesperson said: “We believe that providing API access to our commercial models, including on Google infrastructure, with industry-standard practices and terms, represents a responsible approach to supporting national security.”

The spokesperson also said the company is strongly committed to the view that AI should not be used for domestic mass surveillance or autonomous weaponry without appropriate human oversight.

The Pentagon has previously said it does not want to use AI to monitor Americans on a mass scale or build weapons that operate entirely without people involved. Still, it has pushed for broad legal access to advanced AI systems.

The deal comes as competition grows among technology firms seeking defence contracts linked to AI.

In 2025, the Pentagon signed agreements worth up to $200 million each with several leading AI companies, including Google, OpenAI and Anthropic.

Anthropic later had some challenges after refusing to remove restrictions tied to autonomous weapons and surveillance. It was reportedly labelled a supply-chain risk.

Google’s decision may also revive internal stresses. More than 560 employees reportedly signed an open letter urging Chief Executive Sundar Pichai to reject military AI work.

The company faced a similar backlash in 2018 during Project Maven, when staff protested Google’s involvement in a Pentagon drone programme. Google later withdrew from that project.

Independent clinics have been patching these problems with more headcount for decades, even as labor costs rise due to a shortage of highly specialised workers and burnout deepens.

TriFetch was built to take that administrative weight off the clinic. By automating the three workflows that dominate clinic operating costs: patient calls, referral processing, and prior authorization, TriFetch helps clinics save on 50%+ of administrative costs while also increasing revenue.

This can translate into 1m+ savings for a mid-size practice.

Today, TriFetch announced a $1.9 million pre-seed round led by Nexus Venture Partners with participation from angels from Google, Hipprocratic, Mercor, MIT to scale its automation platform for the front, mid, and back office of specialty care.

The pressure TriFetch is targeting is structural. Independent clinics face the same administrative load as large health systems with a fraction of the staff.

A single prior authorization can take 45 minutes and referral coordination often means hours on hold. Patient calls pile up while the front desk triages between the waiting room and the phone line and those left unanswered go to an inbox with hundreds of unread messages.

“Clinics are doing everything they can to keep up, but the administrative workload keeps expanding,” said Varuni Sarwal, CEO and co-founder of TriFetch. “We built TriFetch to plug into how clinics already run and take the tasks staff dread the most off their plate, calls, referrals, and prior auth, so teams can focus on the parts of care that require the human touch.”

TriFetch automates the three workflows that eat the most staff time. Its multilingual voice agent handles patient calls end to end – inbound inquiries, outbound scheduling, and follow-ups.

Its referral engine routes and processes referrals, verifies eligibility, and books patients with humans in the loop.

And its prior auth automation submits and tracks requests so paperwork delays never push costs onto patients. The platform plugs directly into how a clinic already runs, with no EHR migration or retraining required.

“Clinics don’t need more software where every new tool adds another tab, another login, another thing to learn; they need less friction. TriFetch integrates as the connective tissue of a clinic’s existing operations, adapting to the clinic’s ecosystem and not the other way around,” says co-founder and COO Rosemary He, who is leading the product team.

Cofounders Varuni Sarwal and Rosemary He met at UCLA while completing their PhDs in Computer Science, where they worked at the intersection of AI and healthcare. Varuni’s research applied machine learning to tabular EHR to predict depression and sepsis while Rosemary built computer vision models to predict Alzheimer’s progression in longitudinal 3D medical images.

After publishing in top venues like Nature and ICML, the contrast they saw  from the inside was hard to ignore: while AI was compounding at the frontier, most specialty clinics still ran on fax, phone trees, and manual paperwork.

They built TriFetch to bring that capability into specialty care, and have taken a forward-deployed approach, embedding alongside clinic teams until the system runs end-to-end.

TriFetch is currently running multiple active pilots with specialty clinics across California. In an ophthalmology clinic, the doctor and his staff are being overtaken by phone calls and prior authorizations while trying to deliver the highest quality of care to his patients. In a cardiology clinic, staff have been overwhelmed by patient inquiries and internal routing needs.

In a GI practice, two staff members work full-time processing up to 100 referrals a day and calling patients to schedule them.

TriFetch handles that workflow end-to-end, freeing roughly 16 hours of staff time a day and returning more than $200,000 a year to the clinic. For a mid-size specialty practice, that range of recovered costs and captured revenue can run anywhere from $500,000 to $1.4 million a year.

Dr. Shashi Ganti, Ophthalmologist, Cal Retina MD added:

“Clinics up and down the US are facing the same administrative headache. Working with TriFetch, we’ve been able to relieve our staff from managing patient calls and scheduling: freeing them up from hundreds of voicemails and phone calls  to focus on the patients in front of them. AI can be incredibly powerful when adopted safely, and I can’t think of a better team to trust with that in my clinic.”

As AI adoption accelerates in healthcare, most tools either target large health systems, solve a single narrow workflow, or are EMR-specific.

TriFetch is the first unified, EMR-agnostic automation layer purpose-built for independent clinics that can be customized across specialties and locations, deploying alongside existing systems, NextGen, eCW, Athena, and others, in weeks, not quarters, reducing the need for clinics to stitch together multiple vendors to keep their operations running.

Jishnu Bhattacharjee, Partner at Nexus, commented:

“Varuni and Rose are deep domain experts in healthcare AI. Healthcare administrative workflows represent one of the largest untapped opportunities for AI, and the Trifetch team is uniquely positioned to unlock it. They combine deep AI capabilities with real-world clinical understanding to build what we believe can become a category-defining company in healthcare AI. We are excited to partner with Trifetch and support them on this journey.”

Calls, referrals, and prior authorization for specialty clinics are the starting point. Over time, TriFetch plans to build the first AI-native operating layer for healthcare systems, expanding nationwide from independent specialty clinics into multi-specialty groups, primary care, and hospital-owned networks, deepening EHR integrations, and connecting the tools clinics already use to the workflows that keep care moving. The goal: less time on paperwork, more time with patients.

TriFetch is guided by a founding cohort of 10+ strategic advisors drawn from operators inside the country’s leading health systems, including former NextGen co-founder Tim Eggena, senior leaders from Sutter Health, Johns Hopkins, Mayo, UW Health, Revere Health, Springfield Clinic, and UChicago Medicine.

The cohort reflects TriFetch’s operator-first approach: the people who have run the workflows TriFetch automates are the same people helping shape the product.

The new markets are Australia, Indonesia, Japan, the Philippines, Singapore, South Korea and Vietnam, with the rollout covering desktop and iOS in all listed countries except Japan, where it is currently limited to desktop users.

Gemini in Chrome first launched in the United States in January, after which Google extended access to India, Canada and New Zealand in March. With this latest expansion, more users can now use the feature directly inside the Chrome browser.

Google has been adding more Gemini tools to Chrome since last year. The assistant appears in a floating window and can help users with tasks while they browse.

Earlier this year, Google also introduced a sidebar version that can answer questions across multiple tabs. It can summarise long pages, compare information from open tabs and respond without users leaving the browser.

The feature also works with several Google services. Users can schedule meetings through Calendar, check places with Maps, and draft or send emails through Gmail.

Google’s Personal Intelligence feature is also available through Gemini in Chrome. It allows users to connect services such as Gmail and Google Photos to receive more personalised responses.

Users can also ask questions about YouTube videos while staying on the same page.

Another tool, Nano Banana 2, lets users edit or transform images on the web using text prompts in the Chrome sidebar.

Some advanced functions are still limited to the United States. Google said its agentic feature, which can control a browser window and complete tasks for users, is still being tested.

That feature is only available to subscribers on the AI Pro and AI Ultra paid plans in the U.S. for now.

At the same time, Google still holds close to 90% of the global search market.

There is a difference, but one that is closing fast. I have noticed it in my own behaviour. When I want a quick answer, I type into Google. When I want to see something, maybe a place, a product, a real experience, I open TikTok.

Search no longer means what it used to

Search used to be as simple as typing a question, scanning links, and choosing what to read.

Of course, that model still exists, but it is no longer the only one.

Today, search means:

• Watching a short video
• Listening to someone explain
• Seeing results in real time

On TikTok and Instagram, people go beyond looking for answers to looking for proof, context and experience.

This is a transition from information to demonstration.

Google still holds the system together

It is important to be clear that Google has not been replaced.

It is still the starting point for billions of queries every day. It indexes the web, organises information, and delivers results at speed. For anything detailed, including health, finance, and research, it is still the most reliable route.

Its strength is structure:

• Ranked sources
• Verified websites
• Depth and coverage

Even among younger users, Google is usually the final step, even when it is not the first.

Why TikTok is pulling people in

When it comes to search, TikTok works differently from Google, it does not present pages but people.

If I search for a restaurant, I do not get a list. I see someone walking into the space, showing the food, reacting in real time and that is surely what I want.

This is why usage is increasing. Nearly half of consumers now use TikTok for search-like behaviour, and younger users rely on it heavily for discovery.

Do not think it is replacing Google, it is more about answering a different need:

• “What does this actually look like?”
• “Is this worth it?”
• “What do people really think?”

TikTok answers those questions faster.

Speed vs depth

At the top of this is a simple trade-off.

TikTok is fast.

• Answers come quickly
• Content is easy to consume
• Little effort is required

Google is deep.

• More detailed information
• Wider range of sources
• Greater reliability

The difference is technical and behavioural. When I am in a hurry, I want clarity. When I need certainty, I want depth.

The trust problem

This is where the conversation becomes more serious.

Google’s results are built on ranking systems that prioritise established sources. TikTok relies on creators and engagement.

That changes how trust is formed.

On TikTok, a video can gain visibility because people interact with it, not because it has been verified. Engagement is not the same as accuracy.

However, many people still trust what they see there. Why? Because it feels human, direct and real.

Why people are changing their habits

The transition shows how people now prefer to learn.

Many users, especially younger ones, are moving towards:

• Visual explanations
• Personal experiences
• Quick, practical answers

Research shows that social platforms are now used as discovery tools at scale, particularly for lifestyle, products and local searches.

This is not about abandoning Google but splitting behaviour across platforms.

What this means for businesses

This change is already affecting how brands operate.

It is not enough to rank on Google these days, visibility now depends on:

• Appearing in short-form video
• Being explained by real people
• Showing, not just telling

Search has become fragmented. One platform no longer owns it.

A restaurant, for example, might still rely on Google for location and reviews. But discovery, the moment someone decides to visit, may now happen on TikTok.

Is this just a phase?

The data shows something more permanent.

Usage of TikTok for search is growing endlessly, and at the same time, Google comes first in overall search share.

This is not a replacement but a redistribution.

People are choosing platforms based on intent:

• Google for accuracy and detail
• TikTok for speed and experience

This is becoming consistent.

Where this leaves the user

We are no longer searching in one place. We move between platforms, usually without thinking about it. A question might start on TikTok, continue on Google, and end with a decision influenced by both.

That changes something fundamental. Search is no longer about finding information but about how that information is presented, who presents it, and how quickly it is understood.

So when doing your search, don’t focus on whether Google is losing, or TikTok is winning. When you need an answer today, focus on whether you trust what is fastest or what is most complete.

The research firm predicts Meta will generate $243.46 billion in net advertising revenue in 2026. That would put it just ahead of Google, which is projected to bring in $239.54 billion over the same period.

Meta’s ad business is expected to expand by 24.1% this year, up from 22.1% in 2025. Google’s growth, by contrast, is forecast to hold steady at 11.9%.

Focusing on automated advertising tools is enhancing Meta’s Advantage+ suite, which has gained traction among advertisers who want quicker campaign setup and better returns on spending. 

That demand is helping the company pull in more marketing budgets at a time when brands are watching costs closely.

“In ⁠surpassing Google, Meta has essentially had many of its core ⁠strategies validated,” said Max Willens, principal analyst at Emarketer.

Added to this, Meta has also expanded its ad footprint in recent years. It introduced advertising on WhatsApp and Threads, opening new inventory for marketers. At the same time, Instagram Reels is competing in the short-video space, where TikTok and YouTube Shorts are already strong.

Meanwhile, Google still earns from a mix of businesses, including subscriptions such as YouTube Premium. That spread provides stability, but it may also slow how quickly its ad revenue grows compared with Meta’s more focused push.

The market is concentrated. Emarketer expects Google, Meta and Amazon to account for 62.3% of global digital ad spending in 2026.

Smaller platforms are likely to feel more pressure if ad budgets tighten. Analysts say companies such as Snap and Pinterest are more exposed when advertisers shift spending towards larger, established platforms.

Emarketer noted that recent court rulings involving Meta and YouTube were not included in its projections and are not expected to significantly change the outlook.

On TikTok, the “For You” page drives the vast majority of viewing time. What appears on screen is usually selected rather than random.

I open my phone for a few minutes, intending to check one thing, then I scroll, one video becomes ten, and ten becomes an hour. It seems like a choice, but rarely is. What I am seeing has been filtered, ranked and placed in front of me for a reason.

This is the difference in how information is delivered today. Social media algorithms work differently now, and we no longer go out to find most content; it comes to us, already arranged.

What the system is doing

At a basic level, these systems track how people behave and use that to decide what to show next. They look at how long you watch something, whether you like it, comment on it, share it, or scroll past it.

Over time, patterns are formed. If you pause on a certain type of video, you will see more of it. If you skip something quickly, it fades away. The system keeps testing, adjusting, and refining. It does not understand meaning in the human sense but recognises patterns.

On Instagram, on TikTok, and on YouTube, the process is similar. Content is ranked, not just published. What trends are what holds attention.

That detail is important because attention, not accuracy or balance, is what these systems are built to reward.

Why it works this way

The answer is not complicated, we can say it’s commercial.

Companies like Meta Platforms and Google make most of their money from advertising. The longer people stay, the more adverts they see. The more engaged they are, the more valuable they become.

So the system is tuned to keep people watching and it’s not for a minute, but for as long as possible.

This impacts what is promoted. Content that triggers a reaction; amusement, anger, or curiosity, tends to perform better. Quiet, balanced or less emotional material usually does not travel as far.

In that environment, the feed is not a neutral stream, but an engineered one.

How your feed becomes your world

The process is gradual, but trust me when I say it is consistent.

When I interact with one type of content, the system offers me similar posts. I engage again, and it narrows further. Over time, my feed becomes more focused, more specific.

Eventually, I am not just seeing content, I am seeing a version of reality that has been developed around my past behaviour.

This is where the idea of a “bubble” becomes real. Opposing views don’t appear often anymore. Certain topics take over because they are repeatedly shown while others seem absent.

The result is that I may feel informed but I am often informed within a boundary I did not set.

Reflection or influence?

There is an argument that these systems simply show what people want. After all, you choose what you click, you decide what to watch.

That is partly true, but it is also incomplete.

What you see repeatedly can affect what you think is normal, popular or important. Repetition has weight. If a certain idea appears often enough, it begins to feel familiar. Sometimes, it begins to feel correct.

So the system does both. It responds to behaviour, and it guides it.

The balance between those two roles is where the debate comes in.

What this means

False information can spread quickly if it keeps people engaged. Outrage can travel faster than calm discussion. Trends can appear larger than they are because they are amplified.

In past years, Facebook has faced trials over how content was promoted during political events. Since then, attention has turned to how recommendation systems more broadly can impact public conversation.

The concern is not limited to what is posted but what is pushed.

Why it is hard to look away

There is also a human side to this.

Unpredictable content keeps people watching. One clip may be dull, the next interesting. That makes it harder to stop. Endless scrolling removes natural breaks, there is no endpoint.

Emotion also has a role to play in this too. Content that makes people laugh, argue or react tends to hold them longer and definitely, the system learns that quickly.

Over time, this creates a loop, the system offers, I respond, it adjusts, I stay.

Attempts to set limits

Regulators are trying to limit these impacts. In Europe, new policies are pushing large platforms to explain how content is recommended and to reduce the spread of harmful material. 

Similar discussions are happening in other regions, with calls for more transparency and accountability.

The challenge is increasing, but change is slow. These systems are complex, and they are important for highly profitable businesses.

Can you go back to controlling what you see?

To a degree, yes.

What you choose to engage with does influence what you see next. Following different accounts, pausing on different topics, and ignoring certain content, these have an effect.

But control is limited. The feed is still filtered, still ranked and you are not seeing everything, just what has been selected.

Where this leaves us

It is easy to assume that what appears on screen shows the world as it is, but it usually does not. It shows what holds attention, determined by past behaviour and commercial goals.

That does not mean there is no choice. It means choice operates within a system that is already structured.

So let’s look beyond the feed influencing what we see because it clearly is.

Let’s focus on whether what we see every day impacts how we think, and how much of that thinking is truly our own.