Chainfire

@martinwoodward @GergelyOrosz You're right, they did. It was a bullshit move and illegal in the EU for them, as it is for you. And I will report you to the relevant authorities, just like I did with Anthropic 🤷‍♂️

@GergelyOrosz Hmm, my impression was that the option was checked by default as you are signing up? We can def do a better job now of surfacing this check box on sign-up now we have the option. Previously we didn’t have the option as we didn’t do any training. Will keep digging in thanks.

Happy to chat, but to clarify it’s Copilot usage data for Free, Pro and Pro+ users that is impacted. Not private data at rest and not users of Copilot Business or Enterprise. We emailed everyone impacted earlier in the week and have an in-product banner to let people know how to opt-out ahead of data collection starting. Full details here if you would like to learn more. github.blog/news-insights/…

@GergelyOrosz Don't want to advocate for this bullshit move by GitHub, but when @Anthropic @claudeai started training on user data last August, the selection box to allow it was indeed checked by default (which is also illegal in the EU, and I reported them for it). So he's not entirely wrong.

Maddening - the VP of DevRel at GitHub is gaslighting me saying all other major AI coding providers opt in users as default to train their models. FALSE. Claude Code does NOT do default opt in. You need to decide. Always! It's EXPLICIT GitHub 👎 x.com/martinwoodward…

If you use GitHub (especially if you pay for it!!) consider doing this *immediately* Settings -> Privacy -> Disallow GitHub to train their models on your code. GitHub opted *everyone* into training. No matter if you pay for the service (like I do). WTH github.com/settings/copil…

A quick update on the infamous EU “ChatControl” 🇪🇺 What a turn of events in EU tech policy: from potential mandatory mass scanning of data (“ChatControl”) → to even voluntary scans losing their legal basis (for now). Just months ago, fears were growing around mandatory scanning of private communications in the EU (incl. pictures and videos). Now, talks between the EU Council (Member States) and the European Parliament have collapsed - and the result is a complete reversal. As of April 3, even voluntary scanning of data by platforms loses its legal basis under EU privacy (ePrivacy & GDPR) rules, as the temporary exemption was not extended. A striking example of how fast EU tech policy can turn - and a big win for European privacy advocates.

🇪🇺🎉 HUGE VICTORY! Thanks to your protests, the EU Parliament voted today to END untargeted mass scanning! 💪 But beware: The final decision will now be made in the trilogue with EU governments. The fight continues! ⚔️ All info: patrick-breyer.de/en/historic-ch…

Voor de Nederlanders: box3eerlijk.petities.nl Belastingen waren hier al niet top, maar de voorgestelde wijzigingen voor Box 3 zijn wel heel ernstig.

This man is wasted on the French

Finally! This could be the much needed transformation of business in the EU, but it all hinges on the implementation.

Some personal news: This is my last day at Android Authority, and it also marks the end of my time posting Android news! I've spent the past 10 years writing about Android, and I've enjoyed every moment of it. What began as a hobby in college — tinkering with custom ROMs and modding my phone — turned into a long career, one totally disconnected from my degree in Biomedical Engineering. I am not a developer by trade or education, so I've had to learn a LOT to report on Android the way that I do. I've had my share of missteps along the way, but thanks to the help from the developer and enthusiast community, I've learned SO MUCH. I hope I've been successful in passing much of that knowledge on to you! Thank you to everyone who has followed my journey. I wouldn't have done this work for very long if nobody was there to read it. Your comments encouraged me to invest countless hours into finding new Android features, all in the hopes of sharing these discoveries with people who love the platform as much as I do! If you're wondering what's next, I'll have more to share later. For now, I've got about a month to take care of personal matters, hit the gym, and prepare for relocation! Thanks for everything! P.S. If you're wondering, "where will I get my Android news from now?!" I highly recommend @AndroidAuth! I'm obviously biased, but the team there is fantastic and the content they produce is solid! Also, do check out the @AndroidFaithful podcast — Ron, Huyen, Jason, and Flo are a joy to listen to each week!

Unsurprisingly, the EU Council voted in favor of ending your online privacy today, including your "private" chats. Last stop: European Parliament. The "voluntary" moniker they use to try to sell this is at best misleading. Surprisingly, the Netherlands did the right thing.

@IAmJerdog congrats mate. How much is that in dog years? :)

@badlogicgames Latest @claudeai Code version has completely deleted entire containing directories when I've asked it to revert some dumb little change it made. I thought it was a fluke at first but it has happened three times now. Yeah this is not going great 😂 Git saved the day as usual.

You know what sucks? When the corp you pay $200/month provides an unstable tool, that will silently reset your configuration. Like re-enabling the todo list tool, which I disable for reasons. This is what a vibe coded product is. Just plain chaos. Every release breaks stuff.

@badlogicgames . @AnthropicAI really is speedrunning driving their customers to the competition this past month. Botched updates, quicker limits, training on your chats by default, responsiveness issues, ...

Guess they updated the inference stack again today. Claude will not respond with anything for 3 minutes. Interrupt and ask what's up: immediate response. Let it try again: hangs. Amazing.

Surprise! @ClaudeAI will start training on your data soon. @AnthropicAI is all about safety, so naturally the checkbox that you allow this is enabled by default. I have opinions about this.

Once again the European Council is going to vote on implementing Chat Control (thanks for bringing it up yet again, Denmark!). A dystopian mass surveillance system which scans all your pictures, texts, and e-mail. As the current expected votes stand, it can still be blocked if Germany decides to oppose, or several countries (representing about 80M people) currently in favor switch sides. Contact your MEPs expressing your concerns with just a few clicks: fightchatcontrol.eu If your country is currently in favor or undecided, you can make a difference! Short video explaining (in German, with subtitles): youtube.com/watch?v=cpGSqr… Many more details here: patrick-breyer.de/en/posts/chat-… This vote will impact your freedom if it passes (and the EP also passes it, which needs a lower % of votes than the EC). If this automated system detects anything that doesn't conform to what the EU thinks is acceptable, you will be reported to the authorities. This is the online equivalent of "mail privacy" that they are taking away from you. Your privacy ends with this vote. Nothing to hide? You think this system will not flag that hilarious yet wildly inappropriate meme you just sent to your bestie? That picture of your kids enjoying the pool you sent to the family group? That risque picture you sent your partner? Now imagine when (history shows that the question is not "if", always "when") the EU wants to take some more draconian or totalitarian measures? You will no longer be able to oppose. Identifying and rounding up dissenters has never been easier. Even if you believe the EU would never do such a thing, this system is just asking to be abused by other malicious actors. A central authority needs only one breach for your data to become public (again, not "if", always "when"). If encryption is weakened, it is weakened for all, a hacker's dream come true. Everyone involved in building or maintaining this system is a risk factor. Of course, proponents will try to sell you on the system being built in a safe way, but the technical impossibility of "safe" mass surveillance undermines every assurance they might give. No amount of good intentions or careful procedures can overcome the fundamental reality that a backdoor is a backdoor. As for businesses, all your trade secrets are now only semi-secret. One wonders what this will do for security certifications. Would foreign corporations even want to do business with you if they know you can't protect your communications? Imagine trying to compete with a company that has support from people in power. Surely, there has never been a politician that corrupt. History shows that laws that can be abused, will be abused. And politicians are absolutely aware of the dangers a system such as this imposes. In a classic display of "rules for thee but not for me", they have exempted themselves and law enforcement. Why would they need to do this if it was perfectly safe? How could they do this if they feel the needs of those they represent are as important as their own? The vote to implement chat control has failed to pass before, but they are closer than ever to passing it now. This while a YouGov poll indicates a supermajority (72%) of the population is directly opposed. How is this representation? Particularly nefarious is their attempt to frame it as a way to protect the children. Child protection organizations and abuse survivors themselves have come out saying it does very little for that, and they need resources for proven methods like victim support services, investigation capacity, and prevention programs instead. And if you ask me, protecting children from growing up in the dystopian surveillance state they want to create is a worthy goal by itself. It's even right there in the name - "Chat Control". Not "Child Safety Act" - but *control*. They're telling us exactly what this is about: controlling what you say, know, and by extension, ultimately, *think*. This is the Orwellian nightmare "1984" come to life; once again those in power need reminding that story was a warning, not an instruction manual. If this seems over the top: as someone who builds systems for a living, I know that worst-case scenarios aren't just possibilities, they're eventualities. We have to assume that any power we create will eventually be misused. We have to evaluate systems on their potential for abuse, not their intentions.

@mweinbach With Claude Code activating new limits tomorrow even on the highest plan it's worth comparing how much complex(!) coding you can really do with each on the appropriate model.

Codex has quickly become my favorite CLI for anything development related and the new IDE extension and features make it now probably the best on the market (need to compare to Claude Code, Cursor, etc) at the price they ask and limits you get

@RKBDI I don't disagree (though I happily use the biggest Claude plan my work pays for), but it does seem somewhat beside the point the author makes

@ChainfireXDA just use local models they're free

AI is a Money Trap wheresyoured.at/ai-is-a-money-… Certainly interesting, maybe true.

In Steve we trust

@BinaryOutlook You're not going to believe this but I was actually messing with mesa 😂

@ChainfireXDA I tried not segfaulting for a while but it got boring so I installed mesa-devel. Problem solved.

"Have you tried not segfaulting?" Yes, thank you Claude. Energy well spent.