Motorola partners with GrapheneOS & De-Googled phones: /e/OS and Jolla - Hacker News (Mar 2, 2026)

A major phone maker just teamed up with the GrapheneOS Foundation—yes, that GrapheneOS—hinting at mainstream hardware built to embrace a hardened, privacy-first Android. Welcome to The Automated Daily, hacker news edition. The podcast created by generative AI. Today is march-2nd-2026. I’m TrendTeller, and in the next few minutes we’ll connect the dots across mobile privacy, enterprise device management, the growing backlash to AI-generated junk, and a few standout stories from the developer and hardware worlds.

Motorola partners with GrapheneOS

Let’s start with mobile security and privacy, because Mobile World Congress is already setting the tone for 2026. Motorola—under Lenovo—announced a long-term partnership with the GrapheneOS Foundation. GrapheneOS has built a reputation in security circles for hardening the Android Open Source Project: tightening app isolation, reducing attack surface, and putting privacy controls front and center. The key point here is not “another Android skin.” This is an OEM saying it wants devices engineered with GrapheneOS compatibility in mind, and it wants joint work on research and new security features. Motorola frames it as combining GrapheneOS engineering with Motorola’s own security experience and Lenovo ThinkShield—so expect the conversation to include enterprise threat models, device integrity, and policy controls, not just consumer privacy slogans. Details are still thin, but the direction is clear: if this collaboration turns into real shipping devices with first-class support, it could lower the barrier for people who want hardened phones without a niche hardware scavenger hunt.

De-Googled phones: /e/OS and Jolla

In the same Motorola bundle, there’s also an enterprise angle that’s less flashy but potentially very practical: Moto Analytics. The pitch is real-time visibility into device fleets—things typical EMM dashboards often don’t make easy to see: app stability, battery health trends, and connectivity performance. If you run thousands of devices, the difference between “users complain” and “we can see the regression starting Tuesday after that app update” is enormous. Motorola is positioning this inside the ThinkShield ecosystem and as something that plugs into existing enterprise setups rather than replacing them. And for everyday privacy, Motorola is expanding its Moto Secure app with something called Private Image Data. The idea: when you take a photo, the phone can automatically strip sensitive metadata—like location and device details—while keeping the image itself unchanged. That’s a small feature with big implications for journalists, activists, enterprise field workers, or frankly anyone who forgets how much information rides along in EXIF. Motorola says it’ll roll out first to “motorola signature” devices over the coming months.

AI slop backlash hits Microsoft

Staying on the theme of “phones that don’t default to Big Tech,” two other stories are worth grouping together: /e/OS and Jolla. /e/OS is the e Foundation’s open-source, de-Googled Android-based ecosystem. It’s not just about removing Google apps; it’s about systematically swapping out the services that quietly phone home. That includes replacing Google-dependent components with alternatives like microG, changing default search to Murena Find, and avoiding Google servers for things like connectivity checks, time syncing, and DNS. Even location is rethought: it can use BeaconDB Location Services alongside GPS. What makes /e/OS interesting is how it tries to stay practical. It aims to remain compatible with mainstream Android apps, ships a curated set of everyday apps, and offers more through its App Lounge. It also adds “rated privacy” for apps—showing tracker counts, permissions, and an easy score—plus an Advanced Privacy widget that can manage tracking, and even options like hiding IP address or geolocation. There’s a Murena Workspace account at the center, offering cloud backup, email, and an end-to-end encrypted directory called Murena Vault—positioned as an Office365-style alternative, with self-hosting options for power users. It even builds in parental controls: filtering, app restrictions, screen-time limits, and a parent-friendly “find my device.” If you want to try it, you can buy a phone that ships with it, use a WebUSB installer in a compatible browser, or manually install builds from GitLab. The project leans hard on “auditable privacy” by being open source—and it points to academic work on Android data sharing as part of that argument.

Self-hosted workplace search with Omni

Then there’s Jolla, which is taking orders for a September 2026 batch of its new Jolla Phone—an independent European Linux smartphone built around community input. The structure is very “small company hardware reality”: a limited batch of 1,000 units, €99 refundable reservation, and a €649 final price with VAT in supported markets across the EU, UK, Norway, and Switzerland. Final payment is scheduled by the end of June 2026, with the ability to cancel anytime. On the device itself: Sailfish OS 5, with Jolla AppSupport for Android apps—plus explicit messaging that you can de-Google further by disabling Android support if you want. Hardware includes 5G dual nano-SIM, 256GB storage with microSD expansion up to 2TB, Wi‑Fi 6, Bluetooth 5.4, NFC, and notably a user-replaceable battery and back cover. There’s also a physical Privacy Switch that can be configured to disable things like the microphone, Bluetooth, or Android apps. Specs are modern midrange: a Dimensity 7100, a 6.36-inch FullHD AMOLED, Sony 50MP wide plus 13MP ultrawide cameras, a 32MP selfie camera, around a 5,500mAh battery, and a fingerprint reader in the power key. There’s even an optional 8GB-to-12GB RAM upgrade for €50—priced like a real bill-of-materials decision rather than a marketing tier. The bigger question, as always, is long-term software cadence and ecosystem fit—but as a European-focused, privacy-forward option, it’s one to watch.

Indie games without big engines

Now, let’s pivot from phone privacy to something messier: the backlash against AI-generated junk—what one manifesto calls “AI slop.” A webpage titled “MICROSLOP” goes straight at Microsoft, accusing the company of flooding the internet with low-quality, synthesized, unverified content. The core claim is that Bing’s AI-generated summaries can hallucinate facts and citations, effectively replacing trustworthy sources with confident nonsense—fake reviews, invented statistics, and references that don’t exist. It also criticizes Copilot’s presence across Microsoft products: AI buttons, suggestions, and overlays that allegedly bloat interfaces and steer users toward machine output. There’s a broader anxiety underneath: the feedback loop problem. If models train on internet text, then generate more low-grade text that gets indexed, then the next generation trains on that, you risk “model collapse”—a gradual erosion of quality where synthetic content crowds out original work. And in a very on-the-nose follow-up story: Microsoft reportedly began filtering the word “Microslop” inside its official Copilot Discord server. Messages containing the term get blocked with a moderation notice, and the cat-and-mouse started immediately—people testing variants like swapping in a zero. Reports say moderation expanded, some users lost privileges, and Microsoft restricted parts of the server, including hiding message history in some channels. You can read this two ways. One: it’s basic community management—don’t let a server turn into a drive-by insult wall. Two: it’s a demonstration of how brand protection and AI skepticism are colliding in public, with Microsoft trying to keep Copilot’s community spaces usable while critics interpret any filtering as dodging legitimate concerns. Either way, it shows how charged the AI conversation remains going into 2026.

The Am386: lawsuits over silicon

On the more constructive side of workplace AI, there’s an open-source project called Omni that’s aiming at a problem most companies actually feel every day: finding information across too many tools. Omni is an Apache 2.0 “Workplace AI Assistant and Search Platform.” The headline features are unified search and an AI chat interface across systems like Google Drive and Gmail, Slack, Confluence, Jira, and more. Technically, it blends classic keyword search—BM25—with semantic search using pgvector, and it keeps the stack centered on Postgres via ParadeDB rather than requiring Elasticsearch plus a separate vector database. Omni also includes an agent that can search connected apps, read documents, and run Python or bash commands in a sandboxed container for analysis. The design choice that will matter to security teams is self-hosting: it’s meant to run entirely on your infrastructure so data doesn’t leave the network. And it emphasizes permission inheritance, meaning the AI layer shouldn’t magically expose documents you don’t already have access to in the source systems. It’s a “bring your own LLM” setup—supporting providers like Anthropic, OpenAI, and Gemini, plus open-weight models via vLLM—so you can tune for cost, privacy, or performance. The architecture is a mix of Rust services, Python orchestration, and a SvelteKit frontend, with connectors isolated in their own lightweight containers. In other words: opinionated, modern, and aimed at teams that want AI search without sending corporate memory into someone else’s cloud by default.

Why we stopped talking to strangers

Let’s drop into developer territory for a moment, because one of today’s most detailed posts is a practical reflection on making indie games without big commercial engines. Game developer Noel Berry explains why he tends to avoid Unity or Unreal for his kind of work. His argument isn’t that big engines are bad; it’s that they can be too much engine for smaller games. You pay in overhead—complex workflows, heavy abstractions—and then you end up rewriting core systems anyway. The engine becomes, in his words, mostly an asset loader and editor UI. He also calls out a non-technical risk: forced updates, shifting business terms, and sudden policy changes that can break projects or make it harder to keep older games buildable. Instead, he prefers smaller tools and open-source-friendly frameworks where he can actually understand the whole stack and fix issues himself. On his own stack, he highlights modern C# and the progress in the .NET ecosystem—performance work, better tooling, and hot reload through dotnet watch. For cross-platform foundation, he’s using SDL3, including its newer GPU abstraction designed to span DirectX, Vulkan, and Metal, and he’s built a C# framework layer on top called Foster. For audio, he still depends on FMOD because collaborators need its advanced tooling, even if it brings occasional compatibility pain. He gets very concrete on pipelines too: load everything up front when you can—he mentions a sub-half-second load time for a pixel-art project—and do on-demand loading for larger games. For tooling, he often writes custom editors and uses Dear ImGui plus C# reflection to expose game objects quickly. And on portability, he points to .NET Native AOT improving the feasibility of shipping C# games on console architectures. It’s a thoughtful reminder that “no engine” rarely means “from scratch.” Often it means choosing simpler building blocks you can own.

One story for the hardware history fans: a deep dive on AMD’s Am386—AMD’s clean-room-compatible 80386 clone—and why it arrived so “late.” The post argues the timeline was driven less by engineering and more by business pressure and courtroom gravity. Earlier in PC history, IBM pushed Intel to license CPUs to second sources like AMD to guarantee supply. But IBM didn’t initially want the 386—partly because it threatened IBM’s minicomputer margins—so Intel had little incentive to extend second-sourcing to that generation. Compaq’s 386 PC forced the market’s hand, IBM eventually responded, and by around 1989 IBM began exercising options to make 386s. AMD, meanwhile, did the engineering work—reportedly about two years to implement compatibility—but then came the long legal war. The post claims Intel spent years and roughly $100 million total between the companies’ efforts to delay AMD, including trademark arguments over “386” and attempts to pressure AMD’s 286 rights. On March 2, 1991, AMD won in arbitration and was cleared to release the Am386, with broader disputes lingering into 1995. The chip itself was competitive: 20 to 40 MHz, SX and DX variants, and the Am386DX-40 offering near entry-level 486SX performance at a lower cost—though the broader 386 platform had limited upgrade headroom. It’s a good case study in how “chip history” is often “contract history.”

And finally, a human story that still feels very tech-adjacent: a piece by Viv Groskop about how we’ve collectively stopped talking to strangers. She describes small encounters—a train conversation with an older woman having a hard day, a gentle exchange with a shy waitress missing home—and then asks what changed in public life. Her point isn’t nostalgia for forced chit-chat. It’s that everyday conversation is a learnable skill, and we’re losing the habit because our environments keep removing the need: phones and headphones, remote work, touchscreen ordering, fewer “third spaces,” and lingering post-pandemic instincts to keep distance. She cites research suggesting people underestimate how much they’ll enjoy talking to strangers and how positively they’ll be perceived. And she shares a practical frame: these tiny interactions aren’t about instant intimacy; they’re about exercising social muscles. She also draws a line between real connection and performative “talking to strangers” content online—where consent and sincerity can get blurry. If you want a takeaway that doesn’t feel preachy: lower the stakes. Start with simple, context-based small talk. And if you don’t want to engage, be clear without guilt—something like, “I can’t talk right now.” It’s a reminder that the social layer of life is infrastructure too, and it weakens when we stop maintaining it.

That’s our run for march-2nd-2026: Motorola’s GrapheneOS partnership and enterprise analytics push, privacy-first mobile alternatives like /e/OS and Jolla, a very public fight over “AI slop,” a self-hosted workplace search stack in Omni, a grounded indie-dev take on life without big engines, a reminder that chip wars are often legal wars, and a small nudge to keep our real-world social skills alive. Links to all stories can be found in the episode notes. See you next time.

Motorola partners with GrapheneOS & De-Googled phones: /e/OS and Jolla - Hacker News (Mar 2, 2026)

Our Sponsors

Topics