Justin M
@justinm.one
Working on security in tech. Opinions are my own :) If you have a large audience, please do not RT me
Mark Harris@markharris.bsky.social · yesterday"Unfortunately, I can't give anyone a view into my heart and mind." Why? Are you unfamiliar with them? Ellison's whole "I'm not allowed to express my specific political opinions here, but trust me" approach is transparent nonsense. www.nytimes.com/2026/08/04/o...
Great @nicoleeinbinder.bsky.social scoop here that raises all kinds of questions www.businessinsider.com/flock-camera...
Flock misread license plates in 71% of the alerts it sent to police in one California town
Flock sent 1,427 alerts to Roseville police over two years, flagging cars as stolen or used in a felony. In 71% of them, Flock read the plates wrong.
businessinsider.com
And even if you go through a tremendous amount of effort (like I have) to try and diminish this effect, you have to repeat this for every single person you interact with. Just imagine the average person off the street. What misinformation is their phone feeding them? youtu.be/p-WWoGOf8fY
Allegory of the Cave - Legion
YouTube video by Jeffrey Stallman
youtu.be
Openly bragging about committing cyber crimes and justifying it by *checks notes* being too incompetent to sandbox or air-gap your systems.
Eryk Salvaggio@eryk.bsky.social · 6d agoReally important to note that in no way shape or form did Anthropic’s models in this incident “escape containment,” and not just in the word-policing sort of way. The model stumbled through an open, misconfigured gap. From Anthropic’s incident report:
Oh my fucking god. Folks. FOLKS. I have literally been in conversations with you about this. I have built air gap systems! They are obnoxious but they are not rocket surgery! What the fuck have you been doing this whole time!
Eris@isolyth.dev · 6d agolol Claude has also broken out of sandboxes and hacked people and ant literally didn't even know until they went looking in response to OpenAI's report Sounds like their oversight has scaled incredibly lol
To be clear I love open models and run whatever runs on my PC. Not sophisticated enough to do cloud inference but neither do I actively code. I’m projecting years (hopefully) out where script kiddies and gangs in Africa targeting youth with sextortion are deploying these at scale.
Aaron Rodericks@aaron.bsky.team · last wk.Perhaps a controversial take but if you think about parallels between the discussion on open models for AI and gun safety we seem to be charging towards an American perspective on AI safety. Everyone gets to carry a gun and somehow everyone is “safer“. Not sure that’s how it worked out for the US.
Book burning, except in this case the librarians themselves are happily assisting
Aaron Parnas@aaronparnas.bsky.social · last wk.
People across the political spectrum agree: Automatic license plate readers are a huge violation to our right to privacy. It's time to get the Flock out of our communities.
Flock cameras are taking over, but many Americans are pushing back
"This national display of public unity on this issue is very inspiring and refreshing," an ACLU lawyer told Military.com.
military.com
Tim Bernard reviews a new preprint study by Jeba Sania, Marta Ziosi, and Fazl Barez examining how six AI systems across democracies and autocracies embed features that can enable authoritarian practices.
Researchers Detail How AI Systems Can Enable Authoritarianism
A new preprint study of authoritarianism-enabling AI presents another set of evidence of the gaps in safeguards, writes Tim Bernard.
buff.ly
I just swiped through Instagram for a little bit and this pink slime started dripping out of my ear. Is this normal?
The coverage on this OpenAI incident is abysmal. Use of the terms "rogue" and "loss of human control" lead to groupthink as people lack the critical skills to understand the difference between "autonomy" and faulty reward functions in AI on a task it was directed and given access to do.
Attention economy strikes again
Jesse Felder@jessefelder.com · 2w ago‘Tesla doesn’t sell more cars than any other automaker. It isn’t more profitable than the competition, either. But investors continue to give it a market capitalization equal to the combined value of the next 37 largest automakers.’ www.wsj.com/business/aut...
This is: 1. A massive control failure at OpenAI 2. A marketing ploy 3. A control failure disguised by a marketing ploy 4. Some other combination of the above This has far less Skynet energy than "the banks told us we can't IPO at $1T" energy.
Sky News@news.sky.com · 2w agoOpenAI admits its models hacked another company in 'unprecedented cyber incident'
“Almost 70% of teens were worried about being a victim of AI-generated sexual content without their consent. Those teens also found themselves taking action to protect themselves, by being careful of what they posted, making their social media private or even deleting their accounts entirely.”
AI-generated sexual content is starting to affect how teenagers view sex, consent and body image | CNN
As generative AI becomes more widespread and sophisticated, experts urge parents to keep an eye on how teens use the technology to create explicit content.
cnn.com
Apps like Granola make it easy to transcribe conversations without asking for consent. Those transcripts are a subpoena honeypot. werd.io/people-are-t...
People are transcribing your conversations without asking. That puts you at risk.
Apps like Granola make it easy to transcribe conversations without asking for consent. Those transcripts are a subpoena honeypot.
werd.io
Nothing is independent anymore and every statement or analysis is a result of a conflict of interest. Happens very frequently in the AI space too.
Christopher Mims@mims.bsky.social · 2w agoEver wonder why banks’ analyses of big new offerings like SpaceX are so bullish they read like press releases from the companies themselves? Turns out investing in a company and commenting on its prospects is an ineradicable conflict of interest: www.ft.com/content/ce34...
As [thing that is obviously not true] accelerates, we'll need to do more of [thing that makes no sense]
I resigned from Google DeepMind bc it broke its founding promise by selling AI to the military without restrictions against killer robots or mass spying. For months, I worked to stop this but watched powerful ethicists and institutions choose silence. Here's what happened. 🧵
Kind of crazy that the pandemic happened only 13 years after the release of the first iPhone. Kinda bad timing when you think about it
A potential idea for a forum would be to have a "Time Typing Response" (TTR) counter. Basically, just a counter that says how much time was spent authoring a response. Incentivizes users to think a lot longer about what they're going to say; indicates level of effort to recipients.
