The whole point of SO was to let experts answer specific questions and build a trusted knowledge-base. Having AI answer questions removes the need for humans to even try answering anything.

These are all great ideas for enterprise (especially training on their internal knowledgebase). Not sure it’s worth their while to have a consumer-facing side any more.

• After 50ft, robot battery runs low, turns around.
• Branch has grown through pipe. Robot gets stuck. Now you have two problems.
• Notice they don’t use it for sewer pipes. Giant rats could take them for joyrides.

Totally understandable.

If scanning to help send traffic to your website, that’s cool. If scanning to generate summaries that won’t send any traffic your way. No bueno.

Ultimately, it should be whatever most benefits users.

Pre.

Was about to say they need a camera for XR overlays, but they push it as a “live-streaming video” platform.

So, yeah. They can go pound sand.

If nginx, here’s an open-source blocker/honeypot: https://github.com/raminf/RoboNope-nginx

If you have it set up to be proxied or hosted by Cloudflare, they have their own solution: https://blog.cloudflare.com/declaring-your-aindependence-block-ai-bots-scrapers-and-crawlers-with-a-single-click/

Absolute horseshit. Bulbs don’t have microphones. If they did, any junior security hacker could sniff out the traffic and post about it for cred.

The article quickly pivots to TP-Link and other devices exposing certificates. That has nothing to do with surveillance and everything to do with incompetent programming. Then it swings over to Matter and makes a bunch of incorrect assertion I don’t even care to correct. Also, all the links are to articles on the same site, every single one of which is easily refutable crap.

Yes, there are privacy tradeoffs with connected devices, but this article is nothing but hot clickbait garbage.

Can do what San Francisco does and dump it all out in international waters 🤦🏻‍♂️

https://pubs.usgs.gov/circ/c1198/chapters/193-206_Disposal.pdf

They’ve already got it mapped out.

ILL is the older system, where you make the request to the library and they go try to find it.

There’s a newer “Link+” service where you can do the search yourself and request delivery to specific branches. Supported by a lot of member libraries, for example: https://sfpl.libanswers.com/faq/103121

In our library system, they don’t put Link+ holds with normal holds. You have to go ask someone to look. We get an email that it’s arrived and we have 3 days to pick it up.

In-app purchase required to get them back out.

Made some 30 of them talking to the app server and all the containers inside Docker.

Now we can ask how they’re all doing and asking application-level questions about records, levels, etc., as well as system level questions like how much RAM the db server is using. Makes for a fun demo. Not sure how useful it will be in production.

Beginning if I like it.

Am working my way through a comedian’s interview podcast. Was a tough slog listening to all the episodes through the pandemic.

My car needed some repair work. Insurance set me up for a rental with Hertz who told me not to pay for bridge tolls with my own car’s transponder. When I take the car back, they tell me I’ll be invoiced later for the tolls. Had 4 toll crossings which ordinarily would come to less than $30 (even less if I had used the transponder).

A month later, the Hertz charges show up: $77 (including ‘processing fee’). Called and complained. They said they’d look into it. Never heard back.

Not using them again.

12K. brought to you by Hollywood Face Makeup and CGI alliance.

On-device AI is the way to go. No privacy leak. Doesn’t have server and networking costs.

This specific use case (looking things up in Start menu and settings) is a good one, since finding out which setting to tweak is a major PITA.

Apple just announced at WWDC embedding Foundation Models on phones. Except they will allow apps to access them and give them custom prompts. This doesn’t go quite as far.

Doesn’t replace MCP, but it’s a crapton more secure and well thought out.

Smart move, handing it off to an independent entity. They probably saw how fast MCP took off and realized few people would adopt it if wasn’t.

Similar to other apps, CoverDrop only provides limited protection on smartphones that are fully compromised by malware, e.g., Pegasus, which can record the screen content and user actions.

A lot of the Javascript attributes used for fingerprinting are used to decide WHAT to render and to cache settings so things work smoothly the next time you come back.

For example, the amount of RAM, your WebGL settings and version, presence of audio, mic, and camera, and screen dimensions are all relevant to a game, a browser-based video-conferencing app, or WebASM based tools like Figma.

And unless you want an app to do a full check each time it returns to foreground, it will likely cache those settings in a local store so it can quickly look it up.

If the app needs to send some of this data to the cloud so the server changes what it sends up, they now also have your IP address, rough reverse IP coordinates via ISP, and time. You can use VPN or Tor to obfuscate IP addresses, but you have to remember to turn that on each time you use the app, and in the case of VPN, to disconnect/reconnect to a random server to semi-randomize your IP (or use Tor, which does this for you).

But to answer the first question, changing or disabling those settings could break a bunch of features, especially Single-Page Applications, those using embedded analytics, or any amount of on-device graphics.