The Wide World of Cyber podcast is back, featuring @thekrebscycle.bsky.social, @stamos.org and yours truly! AUDIO: risky.biz/WWC10/ VIDEO: www.youtube.com/watch?v=C8B7...

I guess AI safety work isn’t just woke censorship.

🎹(80’s synthesizer)🎹 🎼Da da da da da da dum da da da Block! Yah Just go ahead and block! Blockkkk!

There’s a super article by @genevievelakier.bsky.social about the “non-First Amendment law of free speech,” which argues, essentially, that it’s a mistake to think the protection of free speech in this country is a matter only (or even principally) of First Amendment doctrine. 1/

I wrote about the Trump Admin's use of federal funding as a jawbone, and why the very reason the First Amendment bar on jawboning is so broad is the very reason jawboning is so effective: intermediaries have strong incentives to just comply. They must not. knightcolumbia.org/blog/federal...

Well, at least taking my grandkids to the museum exhibit on the new listening devices the Chinese invented to have the Qataris plant will be educational for them.

This was a fun mystery to explore with PJ. Have a listen if you are looking for a podcast not about… (gestures around at Rome burning).

Here's my blogpost in support of Krebs. It's different enough from those by @stamos.org or @k8em0.bsky.social that it's worth writing, though the conclusion is much the same. cybersect.substack.com/p/i-am-speak...

Working with Chris Krebs and getting to know his wife Emily and their family has been one of the great honors of my life. The attack against their family should be condemned by all. I have written a post about what is happening to them here: www.linkedin.com/posts/alexst...

First day of class and I have the perfect lecture hall for Trust and Safety.

📢 Excited to share that I started a new company, Corridor, with Ashwin Ramaswami! Corridor is the AI-powered secure by design platform – and we're backed by @stamos.org, Chris Krebs, Christina Cacioppo, @alip.bsky.social at Neo, and Sarah Guo at Conviction. forms.gle/3LsFxtNqzok2...

So I am getting the distinct feeling that Google (in addition to Apple) got a Technical Capability Notice from the U.K.

A New Wide World of Cyber podcast with @stamos.org and @thekrebscycle.bsky.social is out now! We talk about AI, DeepSeek, and regulation. Everyone’s freaking out over this new model. But should they be? VIDEO: www.youtube.com/watch?v=aniE... AUDIO: risky.biz/WWC7/

Following up on the recent news about the U.K. government attempting to force Apple to backdoor its cloud encryption, @stamos.org and I wrote this op-ed in the WSJ. www.wsj.com/opinion/u-k-...

1) This is bad bad bad 2) I'll have more coming out on this soon 3) A huge scoop for @joemenn.bsky.social www.washingtonpost.com/technology/2...

I laughed.

The multiple jailbreaks available to get DeepSeek to eventually figure out something happened in Tiananmen in 1989 are fascinating, and hint towards a Chinese AI censorship regime that is very specific and fragile. Do they just throw the most obvious queries at the model and call it a day?

Deepseek-R1... 1) Is very impressive 2) The 32B version runs very well locally on a 4090 3) Will put a lot of pressure on the big US labs to open-source 4) Will be used in a lot of abuse/spam 5) Has some interesting holes in its knowledge:

This is basically what the DPRK’s teams live for, except with the added frisson of possibly restarting the Korean War* while also racking up the biggest crypto heist in history. *Technically isn’t over yet!

If you want to see how well the war against bots is going on X, just mention that you have a Facebook account that was stolen.

I’m watching the NVIDIA keynote to see if I’ll need to sell a car or my house to afford a 5090.

It’s amazingly considerate of the Iranians/Aliens to adopt the FAA’s approved lighting scheme.

Smartest model ever, still doesn't know who Jonathan Turley is. 🤔

New Bingo card entry dropped.

The @bsky.app team has days to get the fake account problem under control before this poses a serious problem to the value of the platform. It’s bad enough that I would recommend putting some temporary brakes on new account creation, such as required no-VOIP SMS, even if that stunts growth.

1) Many underestimate the investment the PRC has put into influencing the US social media sphere. 2) The destruction of Trust and Safety at X has made it a playground for PRC actors. 3) It's nice to see them whine about BlueSky. 4) But there is nothing magical about BlueSky to keep them away.

Excited to see this out in the world! In which @genevievelakier.bsky.social and I try to make sense of the equivocal signals sent by the Court in last Term's platform cases, and what it might mean for the future of internet regulation. harvardlawreview.org/print/vol-13...

[cw: AI-generated CSAM discussion] New Publication: In a case study for the Partnership on AI, my HAI colleague Caroline Meinhardt & I explain why direct disclosure mechanisms (e.g. content labels) can’t serve as a panacea for all the harms of synthetic media. partnershiponai.org/hai-research...

Listening to today’s hearing on AI harms, I hear Sen. Klobuchar imply that there is not already a federal law against making pornographic deepfakes of kids (there is) and that platforms don't currently have any obligations to do anything about such material (they do). I am going to become The Joker.

Periodical shameless plug 😎 blue.mackuba.eu/skythread/ → Reddit-like thread UI, can also look up quotes blue.mackuba.eu/directory/ → catalog of custom handles & PDSes blue.mackuba.eu/scanner/ → check moderation labels blue.mackuba.eu/labellers/ → labelers mackuba.eu/2024/02/21/b... → Bsky intro

I need to write a BlueSky moderation tool that automatically blocks for content-free quote-tweet dunks. Unicorn startup right there.

YOU GUYS our little baby book made the Kirkus list of best books for 2024!!!! Me and @rmac.bsky.social are freaking out. Thank you so much for supporting Character Limit www.kirkusreviews.com/book-lists/b...

Somewhere in a basement in Beijing, a beleaguered group of MSS analysts are struggling with how to assemble the Chairman’s morning briefing so they can get through it without being executed for pulling a prank.

Me deciding to start posting on Bluesky after getting tired of Threads’ inscrutable algo and gas leak normie replies.

/me checks the headlines I’m either dead and in purgatory or we are all living in a simulation.

And yet, still specifying that CSAM in private channels is totally fine by them

As a follow-up to our work on computer-generated CSAM, we took a closer look at the training data used to train various generative models—most prominently, Stable Diffusion 1.5—to see to what degree CSAM itself might be present in the training data.

On this week's Moderated Content, @stamos.org and I ask: are we The Enemies? We also discuss the struggle to moderate content in the Middle East, the abuse of the EU DSA and SCOTUS taking yet another 1A & platforms case. podcasts.apple.com/us/podcast/m...

You can find my postings at: https://cybervillains.com/@alex

My comment on Jordan’s “Weaponization of CISA” report: “This report grossly misrepresents my work and the work of the MDM subcommittee — a volunteer group that served an advisory, not an operational role for CISA. …”

Hi, are we talking about AI risks this morning? Here’s a NYT story about my colleague @det.bsky.social’s new report about AI-generated CSAM. (Big-ass CW, of course.) https://www.nytimes.com/2023/06/24/business/ai-generated-explicit-images.html

In the course of conducting a large investigation into online child exploitation, SIO discovered serious failings with the child protection systems at Twitter. 1/6 https://www.wsj.com/articles/twitter-missed-dozens-of-known-images-of-child-sexual-abuse-material-researchers-say-58d44f7b