Cloning ethics meets venture capital & Instagram Plus and anonymous Stories - Tech News (Mar 31, 2026)

We’ll start with that biotech story, because it’s the kind of headline that forces a reality check. R3 Bio, a previously quiet startup in California, is now in the spotlight after reporting said it’s pursuing “organ sacks” in monkeys to reduce animal testing—and that its founder also floated far more controversial concepts in private materials, including the notion of non-sentient human bodies as organ sources. There’s no evidence of human cloning here, and the company broadly denies any intent to create human clones. But the bigger takeaway is that extreme life-extension narratives are attracting money and attention faster than ethical frameworks are catching up.

Meta is testing a paid subscription called Instagram Plus in a few markets, and one feature in particular is turning heads: the ability to watch someone’s Story without them knowing. Add-ons reportedly include more ways to segment your audience, keep Stories around longer than the standard window, and get richer Story analytics. The business logic is straightforward—subscriptions are predictable revenue, and power users will pay for more control. The societal question is messier: anonymous viewing nudges Instagram toward more surveillance-like behavior, and it’s hard to see how that improves trust between people posting and people watching.

Apple, meanwhile, looks like it’s leaning into a platform strategy for AI rather than trying to out-muscle the leading model makers head-on. Reporting says iOS 27 could introduce “Extensions” that let Siri route requests to third-party AI chatbots, potentially through an App Store-style marketplace for AI add-ons. If that happens, Siri becomes less of a single assistant and more of a traffic director—giving users choice while letting Apple keep its grip on distribution. But Apple is also drawing clearer lines: it reportedly removed a “vibe coding” app called Anything from the App Store, citing rules against apps that can download or execute new code in ways that change functionality outside review. That tension—opening the door for AI services while policing AI-driven app-building—will define a lot of the iPhone’s next phase.

Over in Europe, Mistral AI just made a big infrastructure move: it secured more than €750 million in loans to expand computing capacity and build a new data center near Paris. This is notable not just for the size, but for what it signals—Europe wants serious, local compute so companies and governments aren’t forced to rely on a small set of U.S. cloud giants. With AI, access to high-end computing has become a strategic choke point. The fight isn’t only about better models; it’s about who can train and run them, where, and under which laws.

On the developer side, OpenAI released an open-source plugin that lets Claude Code users invoke Codex inside an existing Claude workflow. In practice, that means a team can bounce between two major coding assistants without constantly context-switching—using Codex for targeted reviews or longer-running tasks while staying in the same working rhythm. It’s another sign that coding tools are becoming ecosystems, not single products, and that vendors increasingly win by fitting into whatever workflow engineers already have.

At the same time, a cluster of new commentary is converging on one message: the bottleneck with coding agents isn’t writing code—it’s controlling outcomes. One essay argues tickets now behave like prompts; if you describe the problem too narrowly, agents can lock onto the wrong scope and reproduce the worst habits of fragmented work, just at machine speed. Another warns that the key metric is verification cost: if it’s expensive to confirm the result is correct, delegating to agents can create “verification debt” that comes back later as instability and outages. And from the security angle, there’s a darker claim: agents could make vulnerability research vastly cheaper and more scalable, collapsing “attention scarcity” and raising the baseline risk for every neglected system that’s hard to patch. Put it together and the message is clear—automation raises the premium on guardrails, tests, and review discipline.

That pressure is showing up in AI economics too. After a long stretch where model prices fell and usage exploded, some analysts think the market could tilt toward premium pricing—where the most capable models become desirable precisely because they’re expensive and confer an advantage. A leaked mention of a rumored, higher-end Claude variant added fuel to that idea. If top-tier intelligence gets significantly pricier, it could widen the gap between companies that can afford “the best brain on tap” and those that can’t—turning compute access and capital into an even bigger moat.

In culture and media, an AI-generated TikTok series called “Fruit Love Island” reportedly rocketed to millions of followers in just days, with episodes pulling in huge view counts almost immediately. It’s a parody reality show starring animated fruit, produced at a pace that would be punishing for human teams. The conversation it’s igniting is less about whether it’s art, and more about what happens when the algorithm rewards volume, novelty, and fast iteration—and when content can be generated almost as quickly as it can be consumed. It also puts a spotlight on the resource footprint of AI-heavy media production, which ultimately runs on data centers and electricity.

Now to security and conflict, where the line between digital and physical defense keeps blurring. Reports say Iran has stepped up cyber operations tied to the current conflict, including mass text-message campaigns aimed at panic and phishing—trying to trick people into installing fake apps that grab personal data. Researchers also described more destructive attacks attributed to Iran-linked actors. Separately, Ukraine says it’s seeing early results from an experiment that allows private companies to form air-defense teams to protect critical sites from drone attacks, integrated with the national command network. Both stories point to the same trend: security is getting decentralized, faster-moving, and more entangled with civilian infrastructure.

Finally, two governance stories that show how rules are being written in real time. California is drafting new procurement standards for AI vendors that want to sell to the state, emphasizing safeguards around public safety and civil rights, including protections related to exploitative and violent content and discrimination risks. In parallel, China announced a new World Data Organisation aimed at shaping data governance and reducing cross-border compliance friction. Different motivations, different politics—but the same underlying reality: whoever sets the norms for data and AI will shape markets, research, and what’s allowed to scale.

And in space, NASA has reset Artemis with a clear shift in tone: less “race to plant a flag,” more “build something that lasts.” The roadmap now emphasizes sustained lunar presence, adds a nearer-term systems test in Earth orbit, and deprioritizes the Gateway station to focus resources on surface infrastructure. It’s a bet that repeatable operations matter more than a single landmark moment—and that long-term presence is how countries ultimately shape the rules of the Moon.