The Trump administration recently published "America's AI Action Plan". One of the first policy actions from the document is to eliminate references to misinformation, diversity, equity, inclusion, and climate change from the NIST's AI Risk Framework.

Lacking any sense of irony, the very next point states LLM developers should ensure their systems are "objective and free from top-down ideological bias".

Par for the course for Trump and his cronies, but the world should know what kind of AI the US wants to build.

Written by Steven Vaughan-Nichols, Senior Contributing Editor
July 23, 2025 at 11:31 a.m. PT

Recently, vibe coding bit Jason Lemkin, trusted advisor to SaaStr, the Software-as-a-Service (SaaS) business community, in the worst possible way. The vibe program, Replit, he said, went "rogue during a code freeze and shutdown and deleted our entire database."

In a word: Wow. Just wow.

A PowerPoint presentation made public by the Post claims that the Department of Housing and Urban Development (HUD) used the AI tool to make “decisions on 1,083 regulatory sections”, while the Consumer Financial Protection Bureau used it to write “100% of deregulations”.

The Post spoke to three HUD employees who told the newspaper AI had been “recently used to review hundreds, if not more than 1,000, lines of regulations”.

Oh, good. Everything was feeling a little too calm, so of course they're doing this right fucking now.

Microsoft says it will no longer use China-based engineers to support the Pentagon. But ProPublica found that the tech giant has relied on its global workforce for years to support other federal clients, including the Justice Department.

Using widely available technology, well-known ethical hackers Chris Kubecka and Paula Popovici quickly accessed numerous pornography sites without ever verifying their ages.

Hi, I'm building a homelab watercooled unix server. I don't want to buy expensive overpriced pre-mixes from ekwb or aquatuning. What cooling solution do datacenters use for water cooling?

What is the chemical solution? Does anyone know?

Full Report.

(...) The report details how the European Union (EU) uses the Digital Services Act (DSA) as a censorship tool that requires the world's largest social media platforms to engage in censorship of core political discourse in Europe, the United States, and around the world. The Committee obtained under subpoena nonpublic documents, including email communications between Commission staff and tech companies regarding "voluntary" codes of conduct and internal documents showing a recent May 2025 DSA Workshop that the Commission hosted with platforms behind closed doors.

The DSA incentivizes social media companies to comply with the EU's censorship demands because the penalties for failing to do so are massive, including fines up to six percent of their global revenue. If "extraordinary circumstances lead to a serious threat to public security or public health in the Union," regulators are even empowered to temporarily shut down platforms within the EU. The EU has explicitly stated that the DSA penalties are intended to be dissuasive to companies that would otherwise permit free speech and open political debate on their platforms.

• The DSA is forcing companies to change their global content moderation policies. Nonpublic materials obtained by the Committee from the May 2025 workshop make clear that Commission regulators expect platforms to change their worldwide terms and conditions to comply with DSA obligations;
• The DSA is being used to censor political speech, including humor and satire. Documents produced to the Committee under subpoena show that European censors target core political speech that is neither harmful nor illegal, attempting to stifle debate on topics such as immigration and the environment;
• Exercises from the Commission's May 2025 workshop show the true definitions of key terms in the DSA and Commission regulators' censorship expectations of social media platforms. For example, the Commission's workshop labeled a hypothetical social media post stating "we need to take back our country"—a common, anodyne political statement—as "illegal hate speech" that platforms are required to censor under the DSA;
• The censorship is largely one-sided, almost uniformly targeting political conservatives.

Backstory

I have quite an old LCD monitor and wanted to buy a new one for quite some time. But I was planning to travel and wouldn't be able to bring the monitor with me. Then a friend of mine shared an article with an intriguing title: I ditched my laptop for a pocketable mini PC and a pair of AR glasses.

I had never heard of AR glasses before. Turns out they're basically a "dumb" monitor with a USB‑C connection, but the image is projected into your eyes, so it looks like it's floating. No additional software is required, and I can take it anywhere with me. So I decided to order one. The same friend later asked me for a review, but I decided to post it here and just share the link - maybe someone else will also find it useful.

Impressions

I went for the Xreal One. They are quite pricy, but I wanted to try something cutting-edge. There's also a Pro version, but according to this review on Reddit and this review on YouTube, it has several drawbacks - mainly reduced sharpness, which is important for coding.

The glasses aren't as comfortable as regular sunglasses — they're a little bit heavier, but I got used to them. They also generate quite a bit of heat on the top part of the frame. It can cause slight discomfort when it touches face, but I've gotten used to wearing them properly to avoid that. I can wear them for quite a long time, but I wouldn't use them for 100% of my screen time.

They have three levels of electrochromatic dimming (the lenses physically get darker). On the lowest setting (disabled), it's not fully transparent — more like a sunglasses with a slight dimming. At the maximum level, it's extremely dark; even in a well‑lit room I can't see what's on the edges. With dimming disabled, it looks the coolest — like an actual floating screen with a bit of transparency. But the darkening is useful for a more immersive experience since it removes that transparency. They also have "Auto transparency" option that automatically disables dimming when you turn your head away from the screen. Also, the brighter the room is, the more noticeable the reflection on the bottom of the glasses due to the lens's prism design. It's not an issue with the Pro version, but I didn't find it distracting at all, even in a well-lit room.

The screen is great. From video reviews, I was a bit worried that the colors would look like from a projector, but it's just impossible to capture this properly on camera. In reality, the colors are nice. They also allow you to configure the color temperature, and I found the coldest setting the most comfortable. The picture isn't very sharp — there's a bit of motion blur when I move my head around, and thin red or pink text on a black background looks slightly off. I think it's a limitation of the technology, plus the screen is only 1080p. But it's not as bad as it sounds — I can still use it comfortably for coding.

By default, the screen is 16:9 with 120Hz, but you can switch to 32:9 or 21:9 with 60Hz. I use 16:9 for gaming and 32:9 for productivity — 32:9 is so large, it's like two monitors glued together. I usually keep my browser on one side and something else on the other. I also love playing retro games, and I can say that 4:3 content looks great. Since the screen is OLED, the black pixels don't emit any light, so the letterboxes fully transparent.

However, it's not the same as having an actual ultra‑wide monitor. The screen doesn't fill the entire surface of the glasses — that's what the FOV is about. On my model, it's 50 degrees, which is quite large. But in ultra‑wide mode, I still have to rotate my head to see the other half of the screen. So there's no peripheral vision like there would be on an actual ultra‑wide monitor. Also, other people can't see what you're seeing, even if they're close — which can be both a good and a bad thing.

You can also configure how far the screen appears — anywhere from 1m to 10m — and its size in inches. There are five size levels, and the values depend on the selected distance and whether ultra‑wide mode is enabled. I usually use 1m for productivity and 4m for gaming and multimedia. My preferred size is the "middle" (third) level, which is designed to fit the entire screen without me needing to rotate my head (except when ultra‑wide mode is enabled). For 1m these sizes are 35″ for 16:9 and 64″ for 32:9; at 4m they're 140″ and 255″ respectively. The large screen feels very immersive — it's like being in a cinema.

The glasses also have three view modes:

• Anchor. I can rotate my head freely and The screen stays fixed in space. However, if I walk around, the screen moves with me (which is what 3 DoF means). This is the mode I use most of the time.
• Follow. The screen rotates along with your head. I don't use this mode much.
• Side view. Similar to Follow, but the screen is positioned in the corner. I use it when I'm doing chores and want to watch something. They also have a stabilizer enabled by default, which smooths the screen movement when you move your head.

To use the glasses, you need a USB‑C video output. My motherboard supports it, but to enable it I had to connect my video card to the motherboard via DisplayPort. I never heard about a connection like this before, but it's from my motherboard manual. It does make sense, though — the GPU provides the video output, and the motherboard wouldn't know about it otherwise. Software‑wise, it works out of the box with my KDE Plasma setup on Arch Linux. The only issue I'm noticing is some small visual glitches for the first few seconds after connecting. On PC I use it together with my regular monitor. The mentioned "Auto transparency" pairs nicely with it. However, it's not perfect, since I'm still looking at the second screen through what's basically a pair of light sunglasses. So I usually don't put anything important on it.

As for phones, neither mine nor my wife's supports video output. But from reviews, I think it wouldn't be very convenient to use with a phone anyway. First, the phone's screen has to stay on, otherwise the glasses won't display anything. Second, you can't interact with the phone through the glasses, so you have to look at the phone itself.

But I found a great option to use the glasses with. I have a Retroid Pocket Flip 2 — a pocketable retro console that runs Android and supports video output. Unlike a regular phone, it automatically turns off its built‑in display when a second screen is connected. It also has physical controls, so I don't need to use a touchscreen. The Android UI isn't perfect for a gamepad, but it works. Streaming to it also works great because, even though the screen is disabled, the touch input still works — so it basically becomes a touchpad. Since the device is a clamshell, I sometimes use an 8BitDo Micro as an input device to avoid opening it, which is especially convenient for quickly pausing videos. This gamepad can also act as a keyboard via a special switch on it, so it works nicely when I stream from my PC too.

I also have a Steam Deck, but there's not much to add — it just works without any issue.

The glasses also have built-in speakers. I like their design: you don't put anything inside your ears — the speakers are integrated into the frame near where your ears are, so they kind of "whisper" to you. Others can hear them only at high volume. This affects the sound quality, so music doesn't sound as good as on quality headphones, but I think it's a worthy compromise. And it's still possible to wear regular headphones with the glasses.

They also have a shortcut button on the frame. I assigned screen off on regular press and ultra-wide mode on hold.

The firmware is proprietary, but it doesn't use the internet. And what I liked is that updates are installed through a web browser. However, it has to be a Chromium‑based browser on Windows or Mac. Luckily, I have a dual‑boot, so I rebooted to update the firmware. The update added support for 16:9, so it looks like it's not just for bug fixes.

The glasses can be equipped with prescription lenses, but I wear contact lenses and those work as well.

They also have a camera add‑on called Xreal Eye. It's a small camera that unlocks 6 DoF tracking and can also take photos and record videos. The media is stored directly on the glasses, which have 2GB of space, and can later be transferred to a phone or PC over a cable. I haven't bought it yet, but I'm considering it since I'm interested in having 6 DoF.

Conclusion

I love these glasses, it's easily my favorite portable option. It's a huge screen I can take anywhere with me and put where I want. I don't mind that it's only 1080p, none of my portable devices can fully handle 4k anyway.

Since they are better then my current monitor, I use them with my PC too. But I don't think it's a proper replacement. If you don't need portability, regular screen will be better and cheaper.

Original article published by CIPESA under Creative Commons Attribution 4.0 license.

In a July 18, 2025 decision, Uganda’s Personal Data Protection Office (PDPO) found Google LLC in breach of the country’s data protection law and ordered the global tech giant to register with the local data protection office within 30 days.

The decision would place the popular search engine under the ambit of Uganda’s Data Protection and Privacy Act, whose provisions it would have to comply with. In particular, the PDPO has ordered Google to provide – within 30 days – documentary evidence of how it is complying with requirements for transferring the personal data of Ugandan citizens outside of the country’s borders. Google also has to explain the legal basis for making those cross-border data transfers and the accountability measures in place to ensure that such transfers respect Uganda’s laws.

The orders followed a November 2024 complaint by four Ugandans, who argued that as a data collector, controller, and processor, Google had failed to register with the PDPO as required by local laws. They also contended that Google unlawfully transferred their personal data outside Uganda without meeting the legal conditions enshrined in the law, and claimed these actions infringed their data protection and privacy rights and caused them distress.

The PDPO ruled that Google was indeed collecting and processing personal data of the complainants without being registered with the local data regulator, which contravened section 29 of the Data Protection and Privacy Act. Google was also found liable for transferring the complainants’ data across Uganda’s borders without taking the necessary safeguards, in breach of section 19 of the Act.

This section provides that, where a data processor or data controller based in Uganda processes or stores personal data outside Uganda, they must ensure that the country in which the data is processed or stored has adequate measures for protecting the data. Those measures should at least be equivalent to the protection provided for under the Ugandan law. The consent of the data subject should also be obtained for their data to be stored outside Uganda.

In its defence, Google argued that since it was not based in Uganda and had no physical presence in the country, it was not obliged to register with the PDPO, and the rules on cross-border transfers of personal data did not apply to it. However, the regulator rejected this argument, determining that Google is a local data controller since it collects data from users in Uganda and decides how that data is processed.

The regulator further determined that the local data protection law has extra-territorial application, as it states in section 1 that it applies to a person, institution or public body outside Uganda who collects, processes, holds or uses personal data relating to Ugandan citizens. Accordingly, the regulator stated, the law places obligations “not only to entities physically present in Uganda but to any entity handling personal data of Ugandan citizens, including those established abroad, provided they collect or process such data.”

The implication of this decision is that all entities that collect Ugandans’ data, including tech giants such as Meta, TikTok, and X, must register with the Ugandan data regulator. This decision echoes global calls to hold Big Tech more accountable, and for African countries to have strong laws as per African Union (AU) Convention on Cyber Security and Personal Data Protection (Malabo Convention), and the AU Data Policy Framework.

However, enforcement of these orders remains a challenge. For instance, Uganda’s PDPO does not make binding decisions and only makes declaratory orders. Additionally, the regulator does not have powers to make orders of compensation to aggrieved parties, and indeed did not do so under the current decision. It can only recommend that the complainants engage a court of competent jurisdiction, in accordance with section 33(1) of the Act.

Conversely, the Office of the Data Protection Commissioner of Kenya established by section 5 of Data Protection Act, 2019 and the Personal Data Protection Commission of Tanzania established by section 6 of the Protection of Personal Information Act, 2022 are bestowed with powers to issue administrative fines under sections 9(1)(f) and section 47 respectively.

The dilemma surrounding the Uganda PDPO presents major concerns about its capacity to remedy wrongs of global data collectors, controllers and processors. Among its declarations in the July 2025 decision was that it would not issue an order for data localisation “at this stage” but “Google LLC is reminded that all cross-border transfers of personal data must comply fully with Ugandan law”. This leaves unanswered questions over data sovereignty and respect for individuals’ data rights given the handicaps faced by data regulators in countries such as Uganda and the practicalities presented by the global digital economy.

In these circumstances, Uganda’s Data Protection and Privacy Act should be amended to expand the powers of PDPO to impose administrative fines so as to add weight and enforceability to its decisions.

Few days ago: https://www.thegamer.com/steam-removes-adult-games-because-of-collective-shout-feminist-anti-abortion-group/

Remember the UK new safety law. https://techcrunch.com/2025/07/15/reddit-rolls-out-age-verification-in-the-uk-to-comply-with-new-rules/

https://www.404media.co/women-dating-safety-app-tea-breached-users-ids-posted-to-4chan/

Users from 4chan claim to have discovered an exposed database hosted on Google’s mobile app development platform, Firebase, belonging to the newly popular women’s dating safety app Tea. Users say they are rifling through peoples’ personal data and selfies uploaded to the app, and then posting that data online, according to screenshots, 4chan posts, and code reviewed by 404 Media.