@WarnerCrocker oh no! Hope you dont have to brave the elements today!

"NASA Goddard Space Flight Center in Maryland is closing 13 campus buildings (including ~100 labs) with extreme haste and with no transparent strategy or benefit to NASA or the Nation.

Tens or hundreds of millions of dollars of taxpayer-funded NASA property and labs are at risk of either being discarded, mishandled, or out-of-commission for significant time periods.

High profile missions (e.g., Roman, Dragonfly) will be immediately harmed or endangered."

https://www.gesta-goddard.org/blog/gestas-summary-of-goddard-building-closures-status
#NASA
1/n

Connections
Puzzle #877
🟪🟪🟪🟪
🟨🟨🟨🟨
🟦🟦🟦🟦
🟩🟩🟩🟩

#SilentSunday #photography #photographie #photografie #lacleman #lakegeneva

This is the script for my national radio report yesterday on the dangers of AI browsers like OpenAI's new "Atlas". As always, there may have been minor wording variations from this script as I presented it live on air.

- - -

So the issues surrounding generative artificial intelligence, LLM -- Large Language Model AI, are the stories that just keep on giving, so long as what you're interested in getting is night after night of terrible dreams keeping you awake. And it's obvious that the firms behind so much of this often misinformation-laden, sometimes dangerous trash, just can't control themselves, because they've bet their corporate existence on pushing this garbage into everyone and everything, no matter how much damage it might do. And of course, they're not interested in taking responsibility for that damage, typically claiming that the outputs from these AI chatbots and other LLM AI systems are First Amendment protected free speech and attempts to control it would disrupt their "Hell Bent for AI" business models.

However, questions we likely need to be asking as a society are whether perhaps some of these business models SHOULD be disrupted, and whether the billionaire CEOs running these firms need to be put on short leashes before they wreck all of our lives in the pursuit of their often misguided dreams. I don't have time right here to get into the technical details, but the latest nightmarish fantasy AI brew attracting attention is OpenAI announcing an AI Browser called "Atlas". Now you may recall that OpenAI is the firm that is being sued by parents claiming that the firm's ChatGPT AI chatbot helped their teenage son commit suicide, which gives some indication of why many people view the ethical standards of the firm and its leadership in a negative light.

So this browser turns out to apparently be based on the Chromium Open Source browser project, something OpenAI reportedly did not mention when they announced the browser. Chromium as we've discussed in the past is the basis for a whole array of browsers including Chrome, Microsoft Edge and others. Anyone can take the Chromium sources and build from them. And what OpenAI has done, according to researchers who have been testing it, is to take the useful Chromium foundation and turn it into an anti-privacy and security horror show.

Keeping in mind that a web browser is the gateway into the vast majority of most people's interactions with the web other than specific mobile apps for specific purposes, imagine taking a web browser and then grafting on the most problematic aspects of ChatGPT. Imagine creating a browser where your AI overseer is watching your personal activates on the web and even wants to take over performing potentially dangerous actions on the web on your behalf -- in YOUR name. That's the so-called agentic AI that is a major thrust of Big Tech's accelerating AI hype campaign.

Well, this seems to be how researchers are describing key aspects of OpenAI's Atlas browser. And there are reportedly all sorts of technical vulnerabilities being found in Atlas involving prompt injections and mishandling of malformed URLs and so on, some of which sound like amateur level design flaws.

So where does this leave us? Various other Big Tech AI firms are likely working on their own versions of AI browsers that could bring similar vulnerabilities to even more users and then leave those users on their own when things go wrong. Of course most people ARE free to choose the browsers that they prefer to use, at least for their personal use. But in the very short span of time since Atlas was announced, so much frankly scary stuff has been discovered by researchers about its implementation, that one might want to think long and hard before inviting that kind of technology into your web browsing.

Because once you give this kind of AI access to your life in these ways, you might find it difficult or even practically impossible to ever really get that AI to ever completely leave you alone again. And that's perhaps the scariest part of all.

- - -

L

@lauren Could you expand on what you mean in the next to last sentence?

@Tamami I love your hiking photos!

Celebrate 1 trillion pages archived on the Wayback Machine… through art 🌐🎨

Internet Archive x Gray Area: Trillionth Webpage Net․Art Commissions features works inspired by memory, preservation & digital culture. In person in San Francisco!

📅 Sat, Nov 15, 2025
🕗 5–8 PM PT
📍 Internet Archive, 300 Funston Ave, San Francisco
🎟️ https://blog.archive.org/2025/10/24/celebrate-1-trillion-web-pages-with-original-net-art-works-internet-archive-x-gray-area/

#Wayback1T

Librarians, help us celebrate 1 trillion web pages preserved by the @internetarchive! 🌐
Use our resource guide, complete with templates, visuals & event ideas, to connect your community to the web’s history.
More ⤵️
https://blog.archive.org/2025/10/07/calling-all-libraries-celebrate-1-trillion-web-pages-archived-with-the-internet-archive/

#Wayback1T #libraries #librarians

@mikecane @cyberlyra @baldur @glecharles Then you don’t want it enough. No one will build it unless someone does (want it enough)

@mikecane @cyberlyra @baldur @glecharles Now that there are a bazillion AI coding tools, you’ve no excuse for not making it yourself. ::ducks::

#SilentSunday #photography #photographie #LakeKatherine

@GossiTheDog Jira now has ai built-in to help write user stories. This is dumbfounding. If there were only one thing AI doesn’t know it’s what your users (and by extension you) want your product to do and what their context is. Oh wait, that’s two things (cue Monty Python)…

@futurebird @neurovagrant @pluralistic This one is going around. I don’t often answer calls from unknown numbers, but I got a text message with the same CTA to verify a purchase (high amount, round number) and request to call a number. I checked my card balances just in case but would not have called that number.

The mental toll of constant vigilance cannot be understated.

#SilentSunday #photography #photographie #chicago #NavyPier

@ChuckMcManis You might have missed your calling

@cathygellis.bsky.social I’m sorry to hear that. I hope everything goes well.

@mikecane @stroughtonsmith @WarnerCrocker weird

@mikecane iOS26 is terrible

Wordle Bot says my skill and luck are below average today, but I guessed it in 1.7 fewer turns than average. Go figure

Wordle 1,549 2/6

⬛⬛🟨⬛⬛
🟩🟩🟩🟩🟩