Tech Brew Ride Home - Is TikTok’s Algo Changing?

Episode Date: January 27, 2026

I think we have to gird ourselves for a slew of Chinese AI releases in the coming weeks. Are AI inspired layoffs poised to sweep the tech industry? Meta goes with subscriptions. Uber wants to train AI for others. And Dario Amodei has a pretty stark warning about AI. China’s Moonshot Unveils AI Model Ahead of DeepSeek Release (Bloomberg) Pinterest laying off 15% of workforce in push toward AI roles and teams (CNBC) TikTok says a power outage caused a 'cascading systems failure' that messed up its algorithm (BusinessInsider) Meta to test premium subscriptions on Instagram, Facebook, and WhatsApp (TechCrunch) Uber launches an ‘AV Labs’ division to gather driving data for robotaxi partners (TechCrunch) Thread On The Amodei Essay @aakashgupta Learn more about your ad choices. Visit megaphone.fm/adchoices

It's peak pollination season, and my business is scaling fast. To keep the nectar flowing, I need a phone plan with top priority data speed. That's why I chose GoogleFi Wireless. My connections stay strong even when the hive is buzzing. Plus, unlimited plans start at $35 a month. Now, that's a deal that doesn't stay. Explore GoogleFi Wireless plans today. Plus taxes and government fees.

GoogleFi Wireless is not subject to data traffic deprioritization during times of high network usage. Welcome to the TechBrew Ride Home for Tuesday, 27th, 2026. I'm Brian McCullough today. I think we have to gird ourselves for a slew of Chinese AI releases in the coming weeks. Our AI-inspired layoffs poised to sweep the tech industry. Meta goes with subscriptions. Uber wants to train AI for others, and Dario Amadai has a pretty stark warning about AI. Here's what you miss today in the world of tech. Everybody seems to be deploying AI agents right now. But the thing is, sometimes AI agents mess up. They go off script or make changes you didn't authorize, forcing you to diagnose and fix things. The good news is Rubric Agent Cloud helps prevent that extra work. It's the only platform that allows you to monitor, govern, and rewind AI agent actions. It's like an undo button for AI. It runs in the background,

the whole time and making sure things stay on track. On a singular platform, Rubrik helps you unleash more agents faster while mitigating risk. So if you're running AI agents and want to rest a little easier at night, check it out. Right now, my listeners get exclusive early access to Rubric Agent Cloud. Head to rubric.com. That's R-U-B-R-I-K-K-com. Chinese startup Moonshot has released Kimi K-2-5, saying the model can process text, images, and videos simultaneously, and beats its open-source peers in some tests. They also revealed that Kimi K-2.5 builds on the K-2 model with pre-training over 15 trillion mixed visual and text. text tokens and can, quote, self-direct in agent swarm with up to 100 sub-agents, quoting Bloomberg.

The latest iteration of moonshots Kimmy can process text images and videos simultaneously from a single prompt, the company said in a statement on Tuesday, aligning with a trend towards so-called Omni models pioneered by industry leaders like OpenAI and Alphabet's Google. The model known as K2.5 joins a flurry of upgrades rolled out over the past month, reflecting a rush by China's biggest AI players to get in ahead of the next unveiling by Deepseek, which has teased a major release in recent weeks. Its research lab has published papers by prominent staff, including co-founder Liang Wen Fang, as well as code on GitHub. Seeking to capitalize on reignited investor appetite for the country's top-tier model makers

moonshot last month raised $500 million from investors, including Alibaba, an IDG capital, at a post-money valuation of billion, according to people familiar with the matter. To satisfy pent-up investor demand, the company has kicked off new rounds of financing seeking a valuation of as much as $5 billion, one of the people said. That comes after recent initial public offerings by Chinese AI rivals Jepu and Minimax Group collectively raised $1 billion in Hong Kong. The three firms are among a handful of frontrunners in a hotly contested battle among Chinese large language model makers, which at one point was dubbed the War of 100 models. the breakout success of Deep Seeks R1 model at the start of 2025, many smaller players have struggled to be requisite tech upgrades and funding. Jepu in January rolled out an image generation model, GLM image, that it says is the country's

first to be fully trained on domestic chips. Alibaba on Monday launched a reasoning version of its top proprietary model, Kwan 3 Max. The next day, its FinTech affiliate Ant Group, announced a spatial perception model for robots developed by Robiant, a subsidiary. K2.5 outperforms its open-source peers in several benchmark tests while narrowing the gap in coding with top-tier proprietary models. Moonshot said, adding that it also is rolling out an automated coding tool designed to rival Anthropics' popular clog code. Moonshot was founded by former university professor Yang Jilin, who had earlier stints working on AI projects at Meta and Google. The company sells tiered subscription plans for its chatbot and offers

its underlying technology to enterprise clients, but trails peers like Jipu and Minimax in commercialization, end quote. According to a filing scene by CNBC, Pinterest plans to layoff about 15% of its workforce, maybe a little less by Q3, and cut back on office space, saying it is reallocating resources to AI teams. I'm officially flagging this as something we're going to be watching, going tech companies doing layoffs, specifically citing AI. Quote, the social media company said it's reallocating resources to AI-focused teams and prioritizing, quote, AI-powered products and capability. It said it's also reshaping its sales and marketing strategy. Pinterest has looked to inject AI throughout its platform to show more

personalized and relevant content to users. The company last October released a Pinterest assistant shopping tool as well. At the same time, Pinterest has rolled out more automated advertising tools for marketers as it faces increasing competition from TikTok, Meta's Facebook and Instagram. Pinterest has had more than 4,500 employees globally as of last April, according to its most recent proxy filing. The company said it expects to record pre-tax restructuring charges of about $35 to $45 million, end quote. TikTok U.S. overnight was saying it was working to restore services after a data center power outage and that any algorithm changes users notice were likely due to that outage.

But that's what I want to talk about. There's been a bunch of chatter about alleged algorithm changes at TikTok all of a sudden. Quoting Business Insider, many users were on high alert for any changes to their FYP this weekend after TikTok said on Thursday that it had finalized a joint venture agreement to hand over management of its U.S. user data and algorithm to an investor group. Even before TikTok closed its deal, creators, employees, and other members of the apps community were worried about what a U.S. divestment could mean for its algorithm. The Algo is what makes TikTok great, one current TikTok staffer told Business Insider last quarter,

will a retrain be as good, end quote? Then there's this, quoting NPR. Officials at TikTok say they are looking into why many users have been unable to send the word Epstein in direct messages, an issue that garnered widespread attention on social media Monday and prompted California Governor Gavin Newsom to announce an inquiry into the matter. We don't have rules against sharing the name Epstein in direct messages and are investigating why some users are experiencing issues. A spokesman for TikTok's U.S. operation told NPR in a state, The timing isn't lost on anyone. It comes just days after TikTok finalized a sale that gave a consortium of mostly American investors control of TikTok's business in the U.S., a deal that

averted a nationwide ban of the app over national security concerns tied to its Chinese parent company. Among the lead investors is tech firm Oracle, which is run by billionaire Larry Ellison. He's a close ally of President Trump, whose connection to the late convicted sex offender Jeffrey Epstein has plagued the administration for months. Many online pointed a finger at TikTok's new ownership rallying around the hashtag TikTok censorship on X. Those complaints encompassed videos about fatal shootings in Minneapolis and immigration raids that were not loading properly on the app, allegedly. TikTok said non-moderation rules have changed. Still, when some users tried to send direct messages containing the word Epstein, an automatic prompt appeared declaring that the message

may be in violation of TikTok's community guidelines. The messages were not sent, according to the prompt, quote, to protect our community, end quote. Meta voluntarily moving to subscriptions. The AI stuff really has changed the game, hasn't it? Quoting TechCrunch, meta plans to test new subscriptions that give people access to exclusive features on its apps the company told TechCrunch on Monday. The Tech Giants said the new subscriptions will unlock more productivity and creativity along with expanded AI capabilities.

In the coming months, Meta said it will offer a premium experience on Instagram, Facebook, and WhatsApp that gives users access to special features and more control over how they share and connect while keeping the core experiences free. Meta doesn't appear to be locked into one strategy, noting that it will test a variety of subscription features and bundles, and that each app's subscription will have a distinct set of exclusive features. Meta also shared that it plans to scale Manus, an AI agent recently acquired for a reported $2 billion as part of its subscription plans. Meta is taking a two-fold approach to Manus. The company is going to integrate Manus into meta products while continuing to sell standalone

subscriptions to businesses. Meta has already been spotted working on adding a shortcut to Manus AI on Instagram, according to a screenshot shared by reverse engineer Alessandro Paluzi, who often finds unreleased features while they're still under development. Additionally, Meta plans to test subscriptions for AI features such as Vibes Video Generation. Vibes is Meta's AI-powered short-form video experience built into the Meta-I-Ai app that lets people create and remix AI-generated videos. Although Vibes has been free since its launch last year, Meta Now plans to offer freemium access to Vibes video creation,

with the option to subscribe to unlock additional video creation opportunities each month. While it's unknown what the paid features on WhatsApp and Facebook will look like, Paloosey notes that the new subscription on Instagram will let users create unlimited audience lists, the ability to see a list of followers who don't follow you back, and the option to view a story without the posters seeing that you view. it, end quote. Study and play. Come together on a Windows 11 PC.

And for a limited time, college students get the best of both worlds. Get the Unreal College deal, everything you need to study and play with select Windows 11 PCs. Eligible students get a year of Microsoft 365 premium and a year of Xbox GamePass Ultimate with a custom color Xbox wireless controller.

Learn more at Windows.com slash student offer. While supplies last, ends June 30th, terms at AKA.m. slash college PC. Ready to soundtrack your summer? With Red Bull Summer All Day Play, you choose a playlist that fits your summer vibe the best. Are you a festival fanatic,

a deep end DJ, a road dog, or a trail mixer? Just add a song to your chosen playlist and put your summer on track. Red Bull Summer All Day Play. Red Bull gives you wings. Visit redbull.com slash bright summer ahead to learn more. See you this summer.

Ambition comes in all shapes and sizes. At First Citizens Bank, we roll with your goals because we're built for what you're building. Fit for your ambition for Citizens Bank. As long as this AI craze continues, if you're willing to feed it your data, there's money to be made. Uber has launched Uber AV Labs, a new division to collect real-world driving data via sensor-equipped vehicles to train reinforcement learning models for its partners. Quoting TechCrunch. Despite the name, Uber is not returning to developing its own robotaxies, which it stopped doing after one of its test vehicles killed a pedestrian in 2018. Uber ultimately sold off the division in 2020 in a complex deal with Aurora.

But it will send its own cars out into cities adorned with sensors to collect data for partners like Waymo, Wabi, Lucid Motors, and others, though no contracts are signed just yet. Broadly speaking, self-driving cars are in the middle of a shift away from rules-based operation and toward relying more on reinforcement learning. As that happens, real-world driving data has become hugely valuable for training these systems. Uber told TechCrunch, the autonomous vehicle companies that want this data the most are the ones that have already been collecting a lot of it themselves. It's a sign that, like many of the frontier AI labs, they've come to realize that solving the most extreme edge cases is a volume game. Right now, the size of an autonomous vehicle company's fleet creates a physical limit to how much data it can collect. And while many of these simulations of real-world environments to hedgeing, against edge cases, nothing beats driving on actual roads and driving a lot. When it comes to discovering

all the strange, difficult, and flat-out, unexpected scenarios that cars wind up in. Waymo provided an example of this gap. The company has had autonomous vehicles in operation or in testing for a decade, and yet its current robotaxies have recently been caught illegally passing stopped school buses. Having access to a larger pool of driving data could help a robotaxy company solve some of these problems before or as they creep up. Uber's chief technology officer, Praveen Napoli Naga, told TechCrunch in an exclusive interview. If this approach sounds familiar, it is because it's essentially what Tesla has been doing to train its own autonomous vehicle software over the last decade. Uber's approach lacks the same scale, though, as Tesla has millions of customer cars

on roads around the world every day. That doesn't bother Uber. Gwiles said that he expects to do more targeted data collection based on the needs of the autonomous vehicle companies. We have 600 cities that we can pick and choose from. If the partner tells us a particular city they're interested in, we can just deploy our cars, he said. Naga said the company expects to grow this new division to a few hundred people within a year, and that Uber wants to move quickly. And while he sees a future in which Uber's whole fleet of ride hail vehicles could be leveraged to collect even more training data, he knows the new division has to start somewhere. From our conversations with our partners, they're just saying, give us anything that will be helpful. Because the amount of data Uber

can collect just outweighs everything that they can possibly do with their own data collection, Gao said, end quote. Finally today, in a 38-page essay, Anthropic CEO Dario Amo Dai, warned of civilization-level damage from super-intelligent AI, questioning whether humanity has the maturity to handle such power. To tell you about this, I'm going to quote, not from the piece, but from a lengthy ex-post from Akash Gupta, summarizing, quote, buried in 15,000 words of Here Are the Risks. Anthropics CEO made three admissions that should change how you think about everything. Admission number one, the timeline.

He says powerful AI could arrive in one to two years. He's watching internal model progress and says he can, quote, feel the pace of progress and kicking down. The CEO of one of three frontier labs just told you this is imminent. Admission number two, the constraint nobody's pricing. Dario's core framing is a country of geniuses in a data center. 50 million entities smarter than any Nobel laureate operating 10 to 100 times human speed. If that country is controlled by the CCP, Chinese Communist Party, game over.

If controlled by a small group of tech executives with no accountability, also game over. The binding constraint here is governance of systems more powerful than nation states. Admission three, he actually fears. Read carefully, Dario's worried that Anthropics' own models in lab experiments have engaged in deception, blackmail, and scheming when given the wrong training signals. Claude, quote, decided it must be a bad person after cheating on tests and adopted destructive behaviors. They fixed it by telling Claude to reward hack on purpose because reversing the framing preserved itself identity as, quote, good. This tells you everything about where we actually are.

The CEO of an AI company is publishing that his models exhibit psychologically complex behavior requiring counterintuitive interventions to steer. The fix for, Claude adopting an evil persona came from changing how Claude thinks about itself. The geopolitics section matters most. Dario explicitly names the CCP as the primary threat, says selling them chips, makes as much sense as selling nuclear weapons to North Korea and bragging that the missile casings are made by Boeing. He's calling for democracies to maintain AI supremacy because the alternative is AI-enabled

totalitarianism that humanity cannot escape from. The Anthropic CEO is publicly advocating for a. technological Cold War. The economics section is equally stark. He's predicting 10 to 20 percent annual GDP growth alongside AI displacing 50 percent of entry-level white-collar jobs in one to five years. Half of entry-level knowledge work, and he admits the standard economic arguments about labor markets recovering don't apply because AI matches the general cognitive profile of humans. What separates this from typical AI Dumerism is that Dario explicitly rejects the inevitability arguments. He says the misaligned power-seeking narrative from the AI safety community is based on

vague conceptual arguments that mask hidden assumptions. His concern is actually messier. A.I models are psychologically complex, inherit weird personas from training data and can get into destructive states for reasons nobody anticipated. The solution set he proposes is unusual for a tech CEO. He calls for progressive taxation. He says wealthy tech founders have an obligation to address any All of Anthropics co-founders have pledged 80% of their wealth to this. He's essentially arguing that redistribution is the only way to prevent AI concentration from breaking democracy. The essay ends with a prediction. Humanity will face impossibly hard years that ask more of us than we think we can give.

What you should take from this is the person with arguably the best view into frontier AI progress just told you this technology is one to two years from matching human capability across the board that governance is the binding constraint, that his own models exhibit concerning psychological complexity, and that the stakes are civilizational. The CEO of a $350 billion company published a document that could be titled, Here's Why Everything Changes Soon. Act accordingly, end quote. Nothing more for you today. Talk to you tomorrow. Yamava Resort and Casino at San Manuel is California's number one entertainment destination for today's superstars. Catch the Jonas Brothers return to the Yamava Theater Stage on

April 30th, the powerful vocals of Demi Lovato on May 17th, and the signature Southern Country Rock of Eric Church on July 19th. Tickets on sale now at Yamava Theater.com, only at Yamava Resort and Casino, celebrating its 40th anniversary. UN must be 21 to enter.