LA Wildfires Mark the Official Start of the Age of AI Disinformation—and Meet Aria, the Sexy AI Robot Companion

From deepfake wildfire videos to to female companion robots and Elon Musk’s data drought claims, this week marks a turning point. Plus an exclusive interview from Syntho.ai’s Head of Engineering.

Advancement #1: The LA Wildfires: The Official Catalyst for the Era of AI Disinformation

This week, as the horrifying wildfires raged of Los Angeles, social media was flooded with videos showing the devastation of what was destroyed. Countless clips were being shared of the Hollywood sign engulfed in flames, homes consumed by fire, and firefighters heroically rescuing animals in the forest were shared widely, sparking waves of shock and sympathy.

But I noticed something at a frequency I hadn’t before— people didn't trust the videos, because a very good amount of them were fake, but many were unaware of this fact. Many people blindly shared the videos, while others went to social media to quickly prove they were AI. One particularly viral fake even featured Taylor Swift calling the wildfires “divine retribution” against the U.S. for supporting Israel’s bombings in Gaza.

Watch video de-brief here:

We've been long warned that AI could make it harder to trust the videos we watch, but I believe this is the most important update for this week and in awhile for one simple reason: the LA Wildfires is the catalyst that officially marks the age of AI disinformation.

My Initial Thoughts: We knew it was coming, but we’re officially here—there’s no good way to regulate it, and nothing promising on the horizon. It’s unsettling to realize that video, which used to be our most trusted form of evidence, can no longer be assumed real. The stakes are high, and the consequences could be massive. For now, the best we can do is learn to think critically and question everything we see, so I want to teach you some of the things to look out for in AI deepfakes since we can no longer trust what we see.

5 Things to Look for When Identifying Deepfakes:

1. Does It Seem Too Dramatic or Perfect?

   If a video feels almost too perfectly timed or too good to be true, ask yourself if it might be engineered to trigger an emotional reaction. Deepfakes are often designed to grab attention with shocking or sensational content that feels larger-than-life.

2. Unnatural Facial Movements
   Watch for subtle glitches, like odd blinking patterns, frozen expressions, or lip movements that don’t perfectly sync with speech. These are telltale signs of AI-generated manipulation.

3. Odd Lighting or Shadows
   Inconsistencies in lighting—such as mismatched shadows on faces or objects—can indicate that a video has been artificially created or altered.

4. Low-Resolution
   Even polished deepfakes can have pixelation or blurriness, especially around the edges of the face, hair, or other intricate details. Look closely for these imperfections, especially during movement

5. Repetition in Viral Content

   If you’re seeing the same clip reposted by multiple sources with varying captions or contexts, it’s worth questioning its authenticity. Deepfakes often gain traction because they’re reshared with sensationalized headlines, amplifying their reach.

Advancement #2: Elon Musk Said We’re Running Out of Training Data

Elon Musk shook things up this week by saying AI companies have "run out of data" to train their models. What he’s talking about is the easily accessible, free stuff—like the public data you can find on the internet. Experts say this supply is running out fast, with some studies claiming it could be gone as soon as 2026.

To fill the gap, Musk says synthetic data—data generated by AI itself—will play a big role. While there are concerns about AI "hallucinating" or creating errors, synthetic data has its advantages, like being customizable and easier to scale. Combining synthetic data with these untouched resources could open new doors for AI, keeping it on track for future advancements. This isn’t just about running out of data—it’s about learning how to use what’s still out there more effectively.

While this all sounds like a major problem, it’s important to note there’s still a lot of untapped data out there, like private or paid-for datasets, niche industry data, and even information hidden in less-accessible data such on the dark web (which of course could cause ethical issues).

My Initial Thoughts: Instead of giving my opinion here, I decided to go straight to the experts. A friend of mine who is working on the ground with synthetic training data. Watch the video de-brief on my Youtube channel.

Advancement #3: CES 2025: From AI Female Companions to Nvidia’s Pocket-Sized Supercomputer

The Consumer Electronics Show is where the coolest tech gadgets and ideas are revealed every year. It’s where laptops with rollable screens from Lenovo, AI-powered appliances, and even high-tech grills make their debut. This year, however, the spotlight shone brightest on two standout AI-driven innovations: a humanoid robot redefining connection and a tiny AI supercomputer that fits on your desk.

Aria: The Sexy Robot Companion

Realbotix introduced Aria, a humanoid robot built to connect with people on a deeper level. Aria’s AI is designed to be a companion, learn your preferences, respond to your emotions, and carry on conversations that feel natural. With lifelike features and a focus on understanding its users, Aria is positioned as a next-generation companion for those seeking more personal interactions with technology. Realbotix is showing how robots could become a bigger part of our lives—not just as tools, but as something closer to partners.

Aria incorporates advanced motor technology for smoother movement and modular components that can be swapped out or upgraded, making her highly customizable. Her advanced vision system, powered by micro-cameras in her eyes, allows her to track movement, maintain eye contact, and identify objects in real time, and has 17 motors in her face alone driving facial expressions. While Aria is primarily aimed at companionship, its capabilities could be useful in places like hospitals or theme parks, where emotional engagement matters. Her starting price is $175,000.

Project Digits: A Supercomputer in a Cube

On the other side of the CES buzz was NVIDIA’s Project Digits, a small but powerful AI supercomputer. Unlike the huge GPUs NVIDIA usually makes for gamers, this little device is for researchers, startups, and universities that want to work with advanced AI but don’t have big budgets.

During the reveal, NVIDIA’s CEO, Jensen Huang, showed off the compact system, explaining how it packs the power of a full AI server into something small enough to sit on a desk. Project Digits makes advanced AI tools more accessible, giving smaller teams the ability to experiment and create without needing expensive hardware.

My Initial Thoughts: What stood out to me most about Aria wasn’t the technology itself—it was the reaction to her online. The comments were... unsettling. Men flooded forums and social media with questions like, “Will it shut up when I tell it to? I’ll take three.”, “Can I upgrade her boobs?” and “why aren’t they answering the question all men want to actually know”, It wasn’t just inappropriate. It was deeply disturbing.

So, what struck me wasn’t the fact that AI companions like Aria exist. It was how these reactions revealed what some men genuinely think about women and companionship. Aria’s design is meant to push the boundaries of human-robot interaction, but instead, it feels like it’s shining a harsh light on society’s darker attitudes. That, to me, is the most alarming takeaway. If this is how many people think about companions, maybe an AI robot companion is for you.

Advancement #4: Speed Round

Sometimes OpenAI’s Model “Thinks” In Chinese

After OpenAI released its reasoning model, o1, users noticed something strange: it sometimes "thought" in languages like Chinese or Persian, even when asked questions in English. Some speculate this could be due to heavy training on datasets rich in Chinese content or the use of Chinese data labeling services. Others argue that the model processes all text as tokens, and may default to languages it finds more efficient for reasoning tasks. This raises important questions about linguistic biases in AI and how models might prioritize certain languages over others, especially in a multilingual world.

OpenAI Quietly Changes Its Policy on AI Bias

OpenAI recently updated its policy documents, quietly removing a line that said AI models “should aim to be politically unbiased by default.” This change comes after criticism from people like Elon Musk and David Sacks, who have accused AI tools like ChatGPT of being biased against conservative viewpoints. Musk blames the training data and what he calls the “wokeness” of tech companies for shaping AI in a way that leans left. By removing the mention of “politically unbiased” AI, OpenAI seems to be stepping back from making promises about neutrality

TikTok Ban Looming, U.S. Leaders Look to Buy It

TikTok is scheduled to be banned in the U.S. as of January 19, one day before the presidential inauguration. That is unless its Chinese owner, ByteDance, sells its U.S. operations. A group of American business leaders are trying to step in with offers to buy TikTok with some big names, including Elon Musk, reportedly in the mix. If the deal doesn’t happen, TikTok could be pulled from app stores and shut down for American users.

OpenAI Now Let’s You Schedule Reminders and Recurring Tasks

OpenAI just launched a beta version of Tasks, a feature that lets you schedule ChatGPT to handle things for you in the future. Whether it’s setting reminders or automating recurring actions, you can now tell ChatGPT what you need and when, and it’ll take care of notifying you at the appropriate time.

Biden Signs AI Executive Order Days Before Leaving Office

With just days left in office, President Joe Biden has announced a new executive order aimed at advancing AI infrastructure in the U.S. The plan allows private companies to build advanced AI facilities on federal sites, aiming to boost economic competitiveness, national security, and clean energy efforts—allegedly without adding costs to American families.