The outputs, side by side

Back from the Wikimedia CEE conference and still buzzing! ✨

The big one: I gave my first-ever lightning talk. I'd spent days building a deck in Canva, and then the tech broke. What showed up on screen was an odd static version with stickers instead. Not the plan! But the room was so supportive, and the feedback afterwards was lovely. It totally made my day. Only five minutes, yet it's already one of my favorite memories.

On day two I joined the WikiWomen lunch. It was quiet but lovely, just sitting around swapping stories with so many passionate, inspiring women. I came out of it recharged and wanting to do more. It reminded me why this community matters, and how much the sisterhood between us means.

I also spent a lot of the event behind the camera for the CEE Youth Social Media group: hundreds of photos, dozens of video clips, and nearly 300 images uploaded to Wikimedia Commons.

Funny thing, though: I was so busy photographing everyone else that I ended up with zero photos of myself. Not one. 😄

Thank you to everyone who made it so special! 💛 #WikimediaCEE #WikiWomen #Wikimedia

Was it really only five minutes? My first-ever lightning talk at the Wikimedia CEE event is already one of my favorite memories. 💛

I spent days making a deck in Canva… and then the tech broke. What appeared was an odd static version with stickers instead! But the room was so supportive, and the lovely feedback afterward absolutely made my day.

I also spent much of the conference behind the camera: hundreds of photos and dozens of video clips for the CEE Youth Social Media group, plus nearly 300 photos uploaded to Wikimedia Commons. The funny part? I was so busy photographing everyone else that I ended up with zero photos of myself. Not one. 😂

Another highlight was the WikiWomen lunch on day two. Just sitting around a table, swapping stories with passionate, inspiring women, left me feeling recharged and ready to do more. It was a quiet, lovely reminder that this community is about more than what we contribute—it’s also about connection and sisterhood.

Five minutes onstage, a lot of time behind a lens, and so many reasons to keep showing up.

**If data is in a TEE, isn't it just as safe as if it were encrypted?**

Not quite. Both promise that outsiders can't read your data, but their security comes from very different places. Encryption's security rests on math; a TEE's rests on engineering.

Standard encryption algorithms like AES emerge from multi-year public processes in which cryptographers around the world try to break them, and they're built on mathematical problems that have been studied for decades. Nobody sensible relies on a brand-new scheme until it has survived that scrutiny. As a result, there's no known shortcut to breaking these algorithms. Finding one would mean overturning basic things the mathematical community understands about the field.

TEEs don't work that way. A TEE is a protected region inside a processor, designed and built by a particular team of engineers. Every implementation has its own quirks and bugs, and those flaws are typically discovered after the hardware has already shipped, not before. Researchers publish new attacks constantly, and every year brings fresh demonstrations of extracting data that was supposed to be sealed inside a TEE. Vendors do patch these flaws as they surface, but each fix closes one hole at a time. No one has yet built a TEE that can't be broken, and there's no reason to expect perfection anytime soon.

In short, encryption is backed by mathematics that has withstood decades of collective attack, while TEEs are backed by engineering that keeps getting patched. A TEE can meaningfully raise the bar for an attacker, but it doesn't offer the same kind of guarantee.

Not quite. A trusted execution environment (TEE) protects data while a computer is using it. Encryption protects data by making it unreadable without a key. Those are different protections, with different assumptions.

With strong, standard encryption, an attacker who steals the ciphertext still faces a mathematical obstacle: recovering the plaintext without the key should be computationally infeasible. Widely used algorithms have survived years of public scrutiny by cryptographers. That isn’t proof they can never be broken, but it provides a mature, extensively tested foundation for trusting them.

A TEE, by contrast, lets a program work with readable data inside an isolated part of a processor. Its promise is that even a compromised operating system or cloud administrator cannot inspect that computation. Keeping that promise depends on a complicated combination of chip design, firmware, software, and configuration.

That creates more opportunities for failure. Researchers have repeatedly found ways to undermine particular TEEs through implementation flaws and side channels—information leaked through things like timing, caches, or memory access patterns. Fixes help, but vulnerabilities can emerge after hardware is deployed, and some require costly mitigations or replacement hardware.

The distinction isn’t simply “math good, engineering bad.” Encryption also needs sound implementations and key management, and those can fail. Nor does every TEE attack affect every product.

But a TEE adds a substantial set of engineering assumptions to your security model. It can meaningfully reduce risk when processing sensitive data, especially on someone else’s infrastructure. It should not be treated as equivalent to leaving that data encrypted with a key the infrastructure never receives.

**Ring's TAKE system, and what it actually changes**

Many of Ring's camera features, like smart alerts and video search, only work through cloud processing. That's a design choice Ring made, and it has a consequence: to deliver those features, Ring has to decrypt your footage while it sits on its servers.

TAKE is how Ring handles the keys that make this possible. When you use a cloud feature, the decryption keys are handed to Ring's services, which run the processing and return the result you asked for. Under TAKE, this gives Ring access to your footage in the cloud for 24 hours.

After that window, the keys are deleted, but only until you open an older clip or use another smart feature. Then they go right back up to the server.

TAKE does use secure enclaves, which is a small step forward because it makes the base key material harder to pull out directly. But the keys still get passed to services, and those services can be changed. So how much protection does the enclave really buy you?

Ring’s TAKE system adds protection for the keys used to decrypt camera footage—but that isn’t the same as keeping footage inaccessible to Ring’s cloud services.

TAKE uses secure enclaves to make the underlying key material harder to extract directly. That’s a useful safeguard. But decryption keys still become available to services that process your video, and the security of those services—and how they can be changed—still matters.

Why give cloud services access at all? Because Ring designed features such as smart alerts and video search around cloud processing. To provide those features, its systems need to decrypt footage. That’s a product architecture choice, not an unavoidable requirement for every camera.

The other important distinction: temporary access isn’t necessarily one-time access. Expiring or deleting a cloud key doesn’t permanently close the door if viewing an older clip or requesting another feature makes keys available again.

The privacy question isn’t just “Are the keys protected?” It’s also “Who can use them, under what conditions, and what prevents that access from expanding?”

# The Smallest Button on Wikipedia

Wikipedia's Thanks feature lets you acknowledge another editor's work with a single click. The editor gets a notification that someone noticed their edit. There's no message to write, no talk page to open, and no need to find the right words.

Editor Chlod uses it regularly. When someone improves an article he looks after before he gets to it himself, he sends them a thanks. It's also his go-to in busy discussions. Replying in a crowded thread notifies everyone in it, so a thanks is the quieter way to show agreement or appreciation. He compares it to nodding at someone across a room.

He's also frank about its shortcomings. Thanks is tucked away in edit histories, and on some interfaces it doesn't appear at all. Many editors don't know it exists, and in his view even those who do don't use it enough. It isn't a structured task, it doesn't reduce reverts, and it doesn't show up in the numbers that go into quarterly reports. It's far from one of the movement's headline features.

Still, Chlod suspects the fix may not be technical. The movement could simply nudge editors to use it more, especially those who work with newcomers.

His enthusiasm comes from the same place as his support for VisualEditor. That tool is about ten years old, and English Wikipedia recently made it the default for newcomers. His point is that when contributing gets cheaper, more people come in. Thanks applies that logic to appreciation. It makes recognition nearly effortless, and that small signal can be what keeps a new editor coming back.

## A small button for a quieter kind of appreciation

Not every contribution to Wikipedia needs a reply. Sometimes, it just needs a nod.

The Thanks feature lets an editor send another editor a notification acknowledging an edit, without composing a message or opening a talk page. It is a small way to say: I saw what you did, and I appreciate it.

For Wikimedia editor Chlod, that simplicity is the appeal. When someone improves an article he watches before he gets around to it, he sends a thanks. In a busy discussion, where a full reply might notify many participants, Thanks offers a quieter way to acknowledge someone’s contribution.

But a feature that asks so little can also be easy to overlook. Chlod points to its limited visibility: the link is tucked away in edit histories and is not visible in some interfaces. Even editors who know it exists, he thinks, can forget to use it.

His proposed remedy is not necessarily a technical overhaul. The movement could simply remind people that Thanks is there and encourage its use, especially among editors working with newcomers.

That outlook also informs his enthusiasm for VisualEditor: lowering the effort required to participate matters. Thanks lowers the effort required to express appreciation.

It is not a structured editing task or a headline feature. Its value is more modest and more personal: making acknowledgment possible when writing a message feels like too much.

Like nodding to someone across a room, it is small—but it can still mean something.

**A Bangladeshi Worker's Lost Savings and a Job in Malaysia That Never Came**

At 22, a young man from Sherpur, Bangladesh, has lost money on an overseas job he never even started. Speaking to Global Voices by phone on condition of anonymity, he described how his plan to work in Malaysia fell apart.

It began around 2024, when a local intermediary told him there was work available in Malaysia. He dealt mostly with this go-between rather than with a recruiting agency directly, a common arrangement for many prospective migrant workers.

From his side, he did everything asked of him. He completed his medical check and handed over his passport. Payments went out in stages: for visa processing, the medical, and other costs. He did not receive receipts for everything he paid.

He was first told the process would take two or three months. It stretched to six or seven. Each time he asked the middleman what was happening, he got a different explanation, never a clear answer.

By the time everything stopped, his family had put in around BDT 500,000 (about USD 4,079). The money came from savings and the sale of land, as well as about BDT 220,000 (roughly USD 1,795) borrowed at interest.

In the end, he said, he was told that Malaysia had closed its labour market to workers, that the matter was out of the intermediary's hands, and that there was nothing more to try.

His family is left with the loss and the interest-bearing debt. He is left with a medical certificate, a missing paper trail, and no job.

For a 22-year-old man from Sherpur, Bangladesh, the promise of a job in Malaysia left his family facing debt before he ever reached his destination.

Speaking to Global Voices by phone, the worker, who asked not to be named, said he began preparing to migrate around 2024 after a local intermediary told him work was available in Malaysia. He dealt mainly with this go-between, rather than directly with a recruiting agency.

He was initially told the process would take two or three months. Instead, it stretched to six or seven. Whenever he asked for an update, he said, the intermediary offered a different explanation without clearly saying what was happening.

Meanwhile, the family paid in instalments for visa processing, a medical examination and other expenses. They used savings, sold land and borrowed about BDT 220,000 at interest. Their total outlay eventually reached approximately BDT 500,000.

By the worker’s account, he had completed the medical examination and handed over his passport. From his perspective, he had done everything required of him.

Then he was told that Malaysia had closed its labour market to workers and that nothing more could be done. That was the explanation relayed to him, rather than something he could independently establish through the intermediary.

He did not receive receipts for every payment, leaving gaps in the documentation of what his family had spent.

His experience illustrates the financial stakes of a migration process in which a prospective worker pays substantial sums while relying on a middleman for information. For this family, the anticipated overseas earnings never materialised, but the borrowing had already happened.