OpenAI Goes Web-Crawling, Meta Watches Every Keystroke

Alex Shannon: I keep going back and forth on this - I think I actually land on the side that this is a good thing.

Sam Hinton: Really? Because I read the same story and I came out the other end deeply uncomfortable.

Alex Shannon: Look, if Meta’s going to build AI systems that understand how humans actually work, don’t they need real human behavior data?

Sam Hinton: Dude, we’re talking about recording every single keystroke their employees make. That’s not just behavior data - that’s surveillance with a fancy AI label slapped on it.

Alex Shannon: But what if that surveillance actually makes the AI better at helping people? What if it leads to breakthroughs we can’t even imagine yet?

Sam Hinton: That’s exactly the kind of thinking that got us into this mess in the first place.

Alex Shannon: You’re listening to Build By AI, I’m Alex Shannon, and we’re clearly not starting this Tuesday morning on the same page.

Sam Hinton: I’m Sam Hinton, and honestly, today’s news has me questioning everything I thought I knew about where this AI thing is headed.

Alex Shannon: We’ve got OpenAI giving its image generator web browsing powers, Sam Altman throwing some serious shade at Anthropic, and a startup you’ve never heard of that just raised forty million dollars.

Sam Hinton: Plus YouTube’s playing defense against celebrity deepfakes and the AI backlash is about to hit politics hard. Let’s dive in.

OpenAI’s updated image generator can now pull information from the web

Alex Shannon: Alright, so let’s start with the big OpenAI news. They’ve rolled out an updated version of their image generator - and this isn’t just another incremental improvement.

Alex Shannon: The new version has what they’re calling ‘thinking’ capabilities and can actually pull information from the web. So we’re talking about an AI that can browse the internet, process that information, and then generate images based on what it finds.

Alex Shannon: This is paired with their ChatGPT Images 2.0 model, which multiple sources are saying is surprisingly good at generating text within images - something that’s historically been a major weakness for these systems.

Sam Hinton: Yeah, that’s a big deal because we’ve been stuck in this weird uncanny valley with AI images for years. You could get a beautiful picture of a cat, but if there was a sign in the background, it would look like alien hieroglyphics.

Sam Hinton: But the web browsing thing - that’s where this gets really interesting. We’re not just talking about generating images from prompts anymore. We’re talking about an AI that can research current events, check facts, understand context.

Alex Shannon: Right, so if I ask it to create an image of, say, the latest fashion trends in Tokyo, it could theoretically browse fashion websites, look at recent street photography, understand what’s actually happening right now, and then generate something informed?

Sam Hinton: Exactly. And that’s either amazing or terrifying, depending on how you look at it. On one hand, we get images that are way more contextually relevant and accurate. On the other hand, we’ve just given an AI system the ability to consume the entire internet and then create visual content based on that.

Alex Shannon: What do you think the implications are for creators? Because if this thing can browse the web and then generate images in specific styles…

Sam Hinton: That’s exactly what I’m worried about. If it can look at a photographer’s Instagram, understand their style, and then generate similar images, what happens to that photographer’s livelihood? We’re not just talking about generic stock photos anymore.

Alex Shannon: But Sam, isn’t that kind of how human creativity works too? We look at other people’s work, we get inspired, we create something new based on what we’ve seen. How is this fundamentally different?

Sam Hinton: Because humans can’t instantly analyze and replicate the styles of a thousand photographers in five seconds. There’s a difference between inspiration and industrial-scale replication.

Alex Shannon: Fair point. But what about the flip side? What if this democratizes high-quality visual content? What if small businesses that couldn’t afford professional photography can now create compelling visuals?

Sam Hinton: I mean, that’s the promise, right? But we’ve heard this story before - technology that’s supposed to level the playing field ends up consolidating power in the hands of whoever controls the platform. In this case, OpenAI.

Sam Hinton: But here’s what I think people are missing - this web browsing capability means these AI systems are going to get exponentially smarter, exponentially faster. Every day, they’re learning from new content, new trends, new information.

Alex Shannon: That’s a really good point. Traditional AI models get trained once on a fixed dataset. But if this thing is constantly pulling new information from the web, it’s like having an AI that never stops learning.

Sam Hinton: Right, and that raises all sorts of questions about bias, misinformation, copyright. If it’s learning from the web in real-time, it’s also learning from all the garbage on the web in real-time.

Alex Shannon: So for regular people using this, what does this mean? Better AI-generated content, sure, but also AI that understands current events and can create images that feel more relevant and timely?

Sam Hinton: Right, but also AI that might be consuming copyrighted content, personal information, private data - all in service of making better pictures. The question is whether we’re okay with that trade-off.

Alex Shannon: And let’s be real about the competitive implications here. If OpenAI can give their image generator real-time web access, that’s a massive advantage over competitors who are stuck with static training data.

Sam Hinton: Absolutely. This could be the feature that makes OpenAI’s image tools the default choice for businesses and creators. Why would you use anything else if this one has access to current information?

Alex Shannon: What I’m really curious about is how they’re handling the web crawling aspect. Are they respecting robots.txt files? Are they paying for access to premium content? Or are they just scraping everything they can get their hands on?

Sam Hinton: That’s the million-dollar question. And the answer probably determines whether this is a legitimate advancement or just another example of tech companies taking what they want without asking.

Alex Shannon: Keep an eye on this because if OpenAI is doing this with image generation, you can bet they’re thinking about applying the same approach to their language models. Web-browsing GPT with real-time information could change everything.

Sam Hinton: Yeah, and that’s when things get really interesting - or really scary, depending on your perspective. An AI that can browse the web, understand current events, and then generate not just images but text, code, analysis - that’s a different category of tool entirely.

Meta will record employees’ keystrokes and use it to train its AI models

Alex Shannon: Now let’s get into the story that got us fired up in the cold open. Early reports suggest that Meta has developed an internal tool that converts mouse movements, button clicks, and yes - employee keystrokes - into data for training AI models.

Alex Shannon: So we’re talking about comprehensive behavioral tracking of their own workforce, all in the name of making their AI systems better. This goes way beyond the typical workplace monitoring we’re used to.

Sam Hinton: Okay, but here’s what people are missing about this story. This isn’t just about privacy - though that’s obviously a huge concern. This is about Meta essentially turning their employees into unwitting test subjects for AI development.

Sam Hinton: Think about what keystroke data reveals. Not just what you’re typing, but how fast you type, how long you pause between words, how you correct mistakes, your work patterns, your stress levels.

Alex Shannon: But Sam, if we want AI systems that can actually understand and replicate human behavior, don’t they need this kind of granular data? I mean, how else do you train an AI to understand the nuances of human computer interaction?

Sam Hinton: Dude, there’s a difference between understanding human behavior and surveilling your own employees. This is like saying we need to record everyone’s private conversations to build better language models.

Alex Shannon: But hold on - isn’t that essentially what we’ve already done? These language models are trained on massive amounts of human text, much of it scraped from the internet without explicit permission from the authors.

Sam Hinton: That’s… actually a really good point. Maybe the difference is that this is their own employees, people who presumably have some expectation of privacy and fair treatment from their employer.

Alex Shannon: Fair point, but what if this data helps them build AI assistants that are genuinely helpful? What if understanding how humans actually work makes the technology more intuitive and accessible?

Sam Hinton: Here’s my concern - once you normalize this level of workplace surveillance for AI training, where does it stop? Do they record video calls next? Monitor email drafts? Track bathroom breaks to understand human productivity patterns?

Alex Shannon: That’s a slippery slope argument, though. Maybe there are ways to collect useful behavioral data while still respecting employee privacy. Anonymization, aggregation, opt-in programs…

Sam Hinton: But the report doesn’t mention any of those safeguards. And let’s be real - Meta employees probably don’t have much choice in this. You work there, you agree to be part of the AI training data. That’s not really consent, that’s coercion.

Alex Shannon: You know what really interests me about this? It suggests that Meta thinks there’s still significant value in understanding human behavior patterns that they can’t get from existing datasets.

Sam Hinton: Right, which makes me wonder - what exactly are they trying to build that requires this level of behavioral insight? This isn’t just about making better chatbots.

Alex Shannon: Maybe they’re working on AI agents that need to interact with computer interfaces the way humans do? Or maybe they’re trying to understand productivity patterns to build better workplace AI tools?

Sam Hinton: Or maybe they’re building AI systems that can impersonate human behavior so convincingly that you can’t tell the difference. Which is either incredibly useful or incredibly dangerous.

Alex Shannon: What strikes me is that this might actually give us insight into how other tech companies are gathering training data. If Meta is doing this internally, what are they doing with external user data?

Sam Hinton: Exactly. And for anyone working in tech right now, this should be a wake-up call. Your employer might be treating you as a data source first and an employee second. That’s a fundamental shift in the employment relationship.

Alex Shannon: The other thing that bothers me is the lack of transparency. We’re only hearing about this through reports, not from Meta directly. If they’re confident this is ethical and beneficial, why not be open about it?

Sam Hinton: Because they know how it looks. ‘We’re spying on our employees to make our AI better’ is not exactly a great PR message, even if there might be legitimate technical reasons for it.

Alex Shannon: What would you want to see if you were a Meta employee? More transparency? Opt-out options? Better data protection policies?

Sam Hinton: All of the above. But honestly? I’d want to know exactly what they’re building with this data. Because understanding the end goal would help me decide whether I’m comfortable contributing to it.

Alex Shannon: If this gets confirmed and becomes public knowledge, I think we’re going to see some serious pushback from employees and probably some legal challenges. Keep watching this space.

Sam Altman throws shade at Anthropic’s cyber model, Mythos: ‘fear-based marketing’

Alex Shannon: Let’s talk about some good old-fashioned AI industry drama. Sam Altman, OpenAI’s CEO, went on a podcast and basically called out Anthropic’s new cybersecurity model called Mythos.

Alex Shannon: Altman accused Anthropic of using ‘fear-based marketing’ and suggested that the product is way less impressive than they’re claiming. This is pretty unusual - we don’t often see this level of direct criticism between the major AI labs.

Sam Hinton: Oh, this is spicy. Sam Altman usually keeps the diplomatic language when talking about competitors. For him to go this direct means either Anthropic really struck a nerve, or OpenAI feels threatened by whatever Mythos actually does.

Sam Hinton: But you know what? He might have a point about fear-based marketing. The cybersecurity space is notorious for selling products based on worst-case scenarios rather than actual capabilities.

Alex Shannon: What’s interesting to me is the timing. Why is Altman choosing now to go after Anthropic? Is this about market positioning, or is there something specific about the cybersecurity angle that bothers him?

Sam Hinton: I think it’s bigger than just one product. Anthropic has been positioning themselves as the ‘safety-first’ AI company, right? They talk a lot about responsible AI, constitutional AI, all that stuff. And here’s Altman essentially saying they’re using fear tactics.

Sam Hinton: That’s a direct attack on Anthropic’s brand positioning. It’s like saying ‘you guys aren’t actually the responsible ones, you’re just fear-mongering to sell products.’

Alex Shannon: But wait, isn’t OpenAI also in the cybersecurity space? Could this just be competitive positioning? Like, ‘don’t buy their cyber AI, ours is better and less scary’?

Sam Hinton: Absolutely. And honestly, that makes me more skeptical of Altman’s criticism. If he’s got a competing product, of course he’s going to say theirs is overhyped.

Alex Shannon: Let me play devil’s advocate here. What if Altman actually has a point? What if Anthropic is overstating the cybersecurity threats to make their solution seem more necessary?

Sam Hinton: It’s possible. But then again, cybersecurity is legitimately scary right now. State-sponsored attacks, AI-powered social engineering, deepfake phishing - there are real threats out there.

Alex Shannon: True, but there’s also a long history of cybersecurity companies exaggerating threats to sell products. Remember how antivirus companies used to terrify people about computer viruses?

Sam Hinton: Yeah, but some of those viruses were actually dangerous. The question is whether Anthropic is being realistic about AI-powered cyber threats or whether they’re manufacturing fear.

Sam Hinton: But here’s what I find fascinating - this public spat tells us that the AI companies are starting to see each other as real competitors, not just collaborators advancing the field. The gloves are coming off.

Alex Shannon: And that competitive pressure could be good for innovation, but it could also lead to more aggressive marketing claims and faster, less careful development.

Sam Hinton: Exactly. When companies are competing on fear and hype rather than actual capabilities, that’s when you get products that overpromise and underdeliver.

Alex Shannon: What’s also interesting is that Altman chose to do this on a podcast rather than through official channels. It feels more personal than professional.

Sam Hinton: Yeah, it’s like he wanted to get this criticism out there without making it an official OpenAI position. Plausible deniability while still throwing the punch.

Alex Shannon: For people trying to figure out which AI tools to use for their businesses, what do you make of this? How do you evaluate competing claims when the CEOs are throwing shade at each other?

Sam Hinton: Good question. I think you ignore the drama and focus on actual capabilities. Test the tools yourself, look for independent benchmarks, talk to people who’ve used them in production. Don’t let Twitter beef influence your tech decisions.

Alex Shannon: But also pay attention to what they’re competing on. If they’re competing on who can scare you more about cybersecurity threats, that tells you something about their approach to product development.

Sam Hinton: Right. I want to see companies competing on actual performance metrics, not on who can paint the most terrifying picture of the future.

Alex Shannon: This feels like the beginning of a much more competitive and probably nastier phase in AI development. Keep an eye on how these companies talk about each other - it tells us a lot about where the real innovation is happening.

Sam Hinton: And where the real insecurities are. Companies that are confident in their technology don’t usually need to trash-talk the competition this publicly.

AI backlash is coming for elections

Alex Shannon: Now let’s shift gears and talk about something that could affect all of us - early reports suggest that Americans have significant concerns about AI, and that backlash is expected to impact elections.

Alex Shannon: We’re heading into a major election cycle, and it sounds like AI anxiety is becoming a real political issue. Not just deepfakes and misinformation, but broader concerns about AI technology itself.

Sam Hinton: This doesn’t surprise me at all. We’ve been talking about AI like it’s this amazing technological advancement, but for a lot of regular people, AI feels threatening. Job displacement, privacy concerns, just general unease about machines getting too smart.

Sam Hinton: And politicians are really good at picking up on those anxieties and turning them into campaign issues. I bet we’re going to see candidates running on AI regulation, AI safety, maybe even AI bans.

Alex Shannon: What worries me about this is that political backlash tends to be blunt and reactionary. Like, instead of thoughtful regulation that encourages innovation while protecting people, we might get legislation that just tries to stop AI development entirely.

Sam Hinton: Yeah, but Alex, maybe we need some of that blunt reaction. The tech industry has been saying ‘trust us, we’ll self-regulate’ for years, and look where that got us. Maybe political pressure is the only thing that actually forces responsible development.

Alex Shannon: I hear you, but think about what happens if the US decides to heavily restrict AI development while China and other countries keep pushing forward. Do we really want to fall behind on what might be the most important technology of our lifetime?

Sam Hinton: That’s the classic tech industry argument though - ‘we have to move fast or we’ll lose to China.’ Meanwhile, people are losing jobs, their data is being harvested, and they have no say in how this technology gets deployed.

Alex Shannon: But there’s some truth to the competitive argument, right? If we slow down AI development here, other countries won’t necessarily follow our lead. They might just see it as an opportunity to pull ahead.

Sam Hinton: Maybe, but we could also set a global standard for responsible AI development. Sometimes being first isn’t as important as being right.

Sam Hinton: I think what we’re seeing is democracy finally catching up to tech development. For the first time, regular voters might actually have a say in how AI gets built and deployed.

Alex Shannon: That’s a good way to think about it. But I worry about the quality of the debate. Are voters getting accurate information about AI capabilities and limitations, or are they just hearing hype and fear-mongering from both sides?

Sam Hinton: Probably the latter, honestly. Most people’s understanding of AI comes from science fiction movies and sensationalized news articles. That’s not a great foundation for policy decisions.

Alex Shannon: The question is whether politicians actually understand the technology well enough to regulate it effectively. Or are we going to get laws written by people who think AI is magic?

Sam Hinton: Probably the latter, honestly. But you know what? Sometimes imperfect regulation is better than no regulation. At least it forces the conversation into the public sphere.

Alex Shannon: What specific issues do you think voters care most about? Is it job displacement? Privacy? Misinformation? Or just a general sense that technology is moving too fast?

Sam Hinton: I think it’s all of those, but the job displacement thing is probably the biggest. People see AI getting better at creative work, analytical work, even manual labor, and they’re rightfully worried about their economic future.

Alex Shannon: And unlike previous waves of automation, AI feels different because it’s affecting white-collar jobs too. It’s not just factory workers this time - it’s lawyers, doctors, writers, programmers.

Sam Hinton: Right, and those are exactly the people who vote consistently and have political influence. When AI starts affecting suburban professionals, that’s when you get real political momentum for change.

Alex Shannon: So what kind of policies might we see? Universal basic income? Retraining programs? Taxes on AI companies? Outright bans on certain applications?

Sam Hinton: I think we’ll see all of those proposals and more. The question is which ones actually make sense and which ones are just political theater.

Alex Shannon: For anyone working in AI or planning to vote in upcoming elections, pay attention to where candidates stand on AI policy. This could reshape the entire industry based on what happens at the ballot box.

Sam Hinton: And honestly, that might not be a bad thing. Having democratic input into how these technologies develop could lead to better outcomes for everyone, not just tech companies and their shareholders.

AI research lab NeoCognition lands $40M seed to build agents that learn like humans

Alex Shannon: Rapid fire time. First up - early reports suggest a startup called NeoCognition just raised forty million dollars in seed funding. They’re building AI agents that can learn like humans and become experts in any domain.

Sam Hinton: Forty million for a seed round? That’s either incredible technology or incredible hype. The fact that they’re promising agents that can master any domain makes me think it’s probably more hype than substance.

Alex Shannon: Founded by an OSU researcher, so there’s at least some academic credibility there. But yeah, ‘learn like humans’ is a pretty bold claim.

Sam Hinton: I’m skeptical, but if they actually crack human-like learning, that forty million will look like pocket change. Keep an eye on what they actually ship versus what they promise.

Alex Shannon: What gets me is the ‘any domain’ part. That sounds like artificial general intelligence territory, which most experts think we’re still years away from.

Sam Hinton: Right, but maybe they mean something more narrow - like agents that can quickly adapt to new domains with minimal training data. That would be impressive without being AGI-level impossible.

Alex Shannon: Either way, forty million in seed funding shows there’s serious investor appetite for human-like AI, even if the technology isn’t quite there yet.

Sam Hinton: Or it shows that investors are getting swept up in AI hype and writing checks before they fully understand what they’re buying. Time will tell which one it is.

Clarifai deletes 3 million photos that OkCupid provided to train facial recognition AI, report says

Alex Shannon: Next - early reports say Clarifai just deleted three million photos that OkCupid provided for training facial recognition AI. This came after an FTC settlement, and apparently the data sharing goes back to 2014.

Sam Hinton: Wait, OkCupid was giving away user photos for AI training without people knowing? That’s exactly the kind of stuff that’s fueling the AI backlash we just talked about.

Alex Shannon: Right, and this was happening for years before anyone found out. Makes you wonder what other dating apps or social media companies have been doing with user photos.

Sam Hinton: The fact that it took an FTC settlement to stop this is wild. How many other companies are still doing this right now? This is probably just the tip of the iceberg.

Alex Shannon: And the timeline is telling - 2014 to now. That means there’s a whole generation of facial recognition AI that was trained on dating app photos without consent.

Sam Hinton: Which explains why dating apps probably feel so creepy now. Your photos weren’t just being shown to potential matches - they were being used to teach AI systems how to recognize faces.

Alex Shannon: The report mentions that OkCupid executives had actually invested in Clarifai, which adds another layer of ethical complexity to this whole situation.

Sam Hinton: So they were literally profiting twice - once from users uploading photos to their dating platform, and again from selling that data to an AI company they had invested in. That’s pretty gross.

YouTube expands its AI likeness detection technology to celebrities

Alex Shannon: YouTube is expanding its AI likeness detection technology to help celebrities find and remove deepfakes. Early reports suggest talent and their representatives can now use these tools to combat unauthorized use of celebrity likenesses.

Sam Hinton: This is actually smart from YouTube. They’re getting ahead of what’s going to be a massive problem. Celebrity deepfakes are about to get really good and really common.

Alex Shannon: What I like about this is it’s proactive rather than reactive. They’re not waiting for celebrities to sue them, they’re giving them tools to police their own likenesses.

Sam Hinton: Yeah, but notice it’s just for celebrities. Regular people getting deepfaked? Good luck with that. It’s protection for the people who can afford lawyers, not for everyone else.

Alex Shannon: That’s a fair criticism, but celebrities probably generate the most viral deepfake content. If you can stop the high-profile stuff, that might reduce the overall problem.

Sam Hinton: Maybe, but it also creates a two-tiered system where your digital likeness is only protected if you’re famous enough. That doesn’t feel like the right long-term solution.

Alex Shannon: You’re right, though I wonder if this is just the first step. Maybe they’ll expand it to regular users once they work out the technical challenges with celebrity detection.

Sam Hinton: I hope so, but I’m not holding my breath. Tech companies have a history of building features for their most valuable users first and never quite getting around to everyone else.

ChatGPT’s new Images 2.0 model is surprisingly good at generating text

Alex Shannon: And circling back to that OpenAI story - multiple sources are confirming that ChatGPT Images 2.0 is surprisingly good at generating text within images, which has historically been a major weakness for AI image generation.

Sam Hinton: This might sound boring, but it’s actually huge. So much of our visual world includes text - signs, logos, documents, interfaces. If AI can finally handle text properly, that opens up a ton of new use cases.

Alex Shannon: Think about it - AI that can generate realistic screenshots, mock up interfaces, create marketing materials with proper text. That’s going to impact a lot of jobs in design and marketing.

Sam Hinton: Exactly. And combined with the web browsing capabilities we talked about earlier, you’ve got an AI that can research current information and then create professional-looking visuals with accurate text. That’s a powerful combination.

Alex Shannon: What really shows how much AI capabilities have evolved is that we’re now surprised when text generation works well. A year ago, we would have been amazed by any readable text in AI images.

Sam Hinton: Right, it’s like we’ve moved from ‘wow, it can sort of make a picture’ to ‘wow, the typography is actually good.’ The bar keeps getting higher and higher.

Alex Shannon: And for anyone doing content creation or marketing, this could be a game-changer. No more needing separate tools for image generation and text overlay - it’s all integrated now.

Sam Hinton: Which is great for productivity, but also means one more skill set that might become less valuable. Graphic design is about to get a lot more automated.

BIGGER PICTURE

Alex Shannon: If you zoom out and look at everything we covered today, there’s this weird tension happening. On one hand, the technology is getting incredibly sophisticated - web browsing AI, human-like learning, perfect text generation.

Sam Hinton: But on the other hand, people are getting more and more uncomfortable with how this technology is being developed and deployed. The Meta keystroke story, the OkCupid photo sharing, the political backlash - it’s all connected.

Alex Shannon: Right, and you’ve got industry leaders like Sam Altman throwing public shade at competitors while these companies are all racing to build more powerful systems. It feels like we’re approaching some kind of inflection point.

Sam Hinton: I think what we’re seeing is the end of the ‘move fast and break things’ era for AI. The technology is getting too powerful and the stakes are getting too high for the public to just trust tech companies to do the right thing.

Alex Shannon: And the competitive dynamics are getting nastier. When you have CEOs publicly calling out competitors for fear-mongering, that tells you the market is getting more cutthroat.

Sam Hinton: Which could be good for innovation, but it could also lead to corners being cut on safety and ethics. When you’re racing to beat your competitors, responsible development can feel like a luxury you can’t afford.

Alex Shannon: The question is whether we can find a middle ground - regulation that protects people without killing innovation, competition that drives progress without toxic behavior, AI development that’s both ambitious and responsible.

Sam Hinton: And whether the public will have any real say in how this plays out. The political backlash story suggests that democracy might finally be catching up to tech development, but it’s not clear if that will lead to better outcomes or just more chaos.

Alex Shannon: What’s interesting is how all these stories connect to the theme of surveillance and data collection. OpenAI’s web crawling, Meta’s keystroke monitoring, OkCupid’s photo sharing - they all involve taking data without explicit permission to train AI systems.

Sam Hinton: Right, and that’s probably what’s driving a lot of the public anxiety. People are realizing that they’ve become unwitting participants in AI development, whether they wanted to be or not.

Alex Shannon: The NeoCognition funding also fits into this pattern. Forty million dollars for a seed round shows just how much money is chasing AI breakthroughs, which creates pressure to deliver results fast.

Sam Hinton: And when there’s that much money involved, ethical considerations can easily take a backseat to profit motives. Which brings us back to the need for external regulation and oversight.

Alex Shannon: But then you have things like YouTube’s celebrity deepfake detection, which shows that some companies are trying to get ahead of potential problems rather than waiting for regulations to force their hand.

Sam Hinton: True, though as we noted, that protection only extends to celebrities. It’s a perfect example of how market incentives shape who gets protected and who doesn’t.

Sam Hinton: That’s the challenge for 2026 and beyond. Because if we get this wrong, we either end up with AI that’s too restricted to be useful, or too powerful to be safe. Neither of those outcomes is good for anyone.

Alex Shannon: And the window for getting it right might be smaller than we think. The technology is advancing so fast that by the time we figure out the right regulatory framework, we might be dealing with completely different challenges.

Sam Hinton: Which is why these public conversations matter so much. The more people understand what’s happening with AI development, the better equipped they are to make informed decisions about what kind of future they want.

Alex Shannon: Even if those decisions happen through imperfect political processes. Democracy might be messy, but it’s better than leaving all these choices up to tech executives and venture capitalists.

OUTRO

Alex Shannon: Alright, that’s our show for today. Thanks for listening to another episode of Build By AI. We’ll be back tomorrow with more news from the AI frontier.

Sam Hinton: And hey, if you found today’s discussion valuable, hit that subscribe button and tell a friend. This stuff affects everyone, so the more people paying attention, the better.

Alex Shannon: Until tomorrow, I’m Alex Shannon.

Sam Hinton: And I’m Sam Hinton. Keep building.