Mor Naaman

Record-high voter turnout so far in Israel. Like, blasting-the-previous-record high. I am not an expert, but (optimistically) this cannot be good for the ruling coalition. Right? Right??!! (Serbia & Brazil recently provided conflicting signals about turnout, dang it). 4 hours till polls close.

Still unheard-of-high as of 6pm (two more hours)

arxiv.orgCheap AI-Assisted Reverse Engineering and What It Will DoGenerative AI is putting the law of reverse engineering (RE) under severe strain. Existing IP and technology laws assume that RE is expensive and time-consuming, and that it requires specialized exper...

I am sure Greenblatt will be less concerned once he learns Mamdani's post was just an awkward gesture in a moment of enthusiasm.

Joel S.@joelhs.bsky.social

Nothing about President Trump saying protesters at his rallies are being paid by Soros or that Jews who vote for Abdul El-Sayed are bad Jews, but time to complain that Mayor Mamdani accurately described Marc Chagall as an Eastern European artist and not specifically as a Jewish artist. Pathetic.

Mor Naaman reposted

James Grimmelmann

@jtlg.bsky.social

NEW PREPRINT: Cheap AI-Assisted Reverse Engineering and What It Will Do (in submission) We argue that AI coding tools make it easy, cheap, and fast to reverse engineer software and hardware—and IP law is not prepared for this challenge. arxiv.org/abs/2610.35633

schneier.comAI and Liability - Schneier on SecurityEarlier this month, a German court ruled that Google is liable for its AI search summaries. Rejecting defenses like “users can check for themselves,” and that they generally know “that information gen...

The second part of that paragraph is even more pointed: "Why hire human writers, lawyers or doctors when AIs are not only cheaper, but also absolve employers whenever they make a mistake?"

Wolfie Christl@wchr.bsky.social

"AI agents are agents of the person or organization that deploys them—and should be treated by the law as such. ... To allow businesses to hide behind the excuse of faulty AI ... would introduce disastrous incentives for corporate misbehavior" www.schneier.com/blog/archive...

22h

Best explanation so far

Melissa Ryan@melissaryan.bsky.social

The only reason anyone is talking about Marissa Mayer or Dazzle is that her marketing team understands how to deploy rage bait, and most of the Internet still can't spot it in the wild.

linkedin.comAI Rinpoche: Teaching AI & Society in Bhutan (Part 1)As some of you know, I am spending my sabbatical year traveling to 6-7 universities around the globe to teach a 3-day course on AI & Society. Where better to start than Bhutan, the remote Buddhist kin...

First report from my sabbatical year, during which I am traveling to 6-7 universities around the globe to teach a three-day course on AI & Society. Where better to start than Bhutan, the remote Buddhist kingdom in the Himalayas? I wrote it up here (part 1). www.linkedin.com/pulse/ai-rin...

... in which you will find out that Bhutan's "Doctor Google" equivalent is now “AI Rinpoche”.

linkedin.comTeaching in Bhutan: From a Quiet Classroom to a Haunted HouseAs some of you know, I am spending my sabbatical year traveling to 6-7 universities around the globe to teach a 3-day course on AI & Society. This is the second post from my first stop, Bhutan, where ...

Second post from Bhutan! This time, about the amazing (but very quiet) students and how I got to know them better. www.linkedin.com/pulse/teachi...

Mor Naaman reposted

David Mimno

@dmimno.bsky.social

Side benefit of this idea: I can direct prospectives to less established faculty they might not be aware of

Mor Naaman@informor.bsky.social

Show proof that you aren't just spamming. Crazy idea: put on your webpage a list of professors that you contact; no more than 10-15 names. Refer to it in your email. Those names better be in an area close to mine. This is an assessment signal of interest/commitment in me AND the research area.

Mor Naaman reposted

I get 5+ emails *a day* from prospective PhD students. I have seen some advice going around about what to write in these emails. Here's my piece of advice: don't waste your time. Or your tokens. #phdadmissions Or, if you really insist...

theatlantic.comThe Man Who Saved the Internet YesterdayOndřej Surý was a relatively anonymous volunteer working on PHP security. Then the internet exploded.

The most "European IT guy" interview you'll ever read. "Well, my wife says that I’m emotionally dead, but I’m not."

Charlie Warzel@cwarzel.bsky.social

70 percent of the internet was vulnerable yesterday to very bad security flaws inside a coding language. Turns out, *one volunteer* is mostly responsible for maintaining that code & helping push updates. Yesterday Ondřej Surý saved the internet. So I called him up www.theatlantic.com/technology/2...

Include in your email ANY kind of a costly signal that makes clear your email to me is not part of a mass-email campaign. And no, that signal is not "I am very interested in your research and my work is relevant". Everyone and their agents can write this email to any professor anywhere.

Instead, show me a costly signal. Show me that your read my website, or my Bluesky posts, and refer to something specific on it that is not easily off a template (ok this is also possible to do with AI at scale but will at least stand out from the current landscape). Or, even better...

I am as surprised as @karenhao.bsky.social to see this but to be precise, the authors point to the nations (China/US) as the empires, and argue that the corporations are entirely controlled by the sovereign nations. This may be correct in China, but not in the US of Trump/GOP/Citizen United.

foreignaffairs.comEmpires of AIIn this essay, Niall Ferguson, one of the world’s foremost experts on imperial history, teams up the former CEO of Google, Eric Schmidt, and the researcher Lisa Klaassen to make the provocative argume...

Full article: www.foreignaffairs.com/china/empire...

convovo.aiAI Ballot 2026We asked GPT and Grok who to vote for in every 2026 race for Congress and governor. GPT picks the Democrat; Grok picks the Republican.

Aren't the bots "supposed" to decline answering these questions? JFC convovo.ai/ballot/ by/via my colleague Tap Parikh

I sense a trend

Ok your turn @mayor.nyc.gov

The New York Times@nytimes.com

Prime Minister Andy Burnham announced a campaign against listening to music or videos without headphones on public transportation. London commuters weighed in on those plans. trib.al/dW7NL6Y

People seem to be upset but his equating of their humanitarian record is likely accurate

Ron Filipkowski@ronfilipkowski.bsky.social

Trump’s special envoy Steve Witkoff to Vladimir Putin today: “You talk so much about Trump being a humanitarian, you are a humanitarian also, sir. And I salute that. I commend you.”

[late to this] interesting work/method and appreciate the fact that they went beyond the US crowd. Also encouraging that people (somewhat) reject this language from AI. Missing a (difficult) in-the-wild study, though: people may respond to such language favorably even if perceiving it negatively.

Jacy Reese Anthis@jacyanthis.bsky.social

When a chatbot uses anthropomorphic companion language like "I'm worried about you" or "You are great, dear", people actually see it as less likable, humanlike, and trustworthy, according to our new simulation study @GoogleResearch with people from the US, UK, India, and Nigeria.

1000%, will give my own wealth equivalent of $158M if he does

Kevin M. Kruse@kevinmkruse.bsky.social

I'm begging @teamtalaricohq.bsky.social to run against Musk. He's an unpopular and creepy white nationalist asshole billionaire whose approval rating in most polls is 20 points underwater. You can mock Paxton for needing to be bailed out *and* mock the puppet master pulling his strings.

Well here it is. Happy to be out another $100 or so bsky.app/profile/theb...

The Bulwark@thebulwark.com

Talarico: "We just found out that [Elon Musk] is almost single-handedly bankrolling this effort to install Ken Paxton in the U.S. Senate and defeat this movement...Elon Musk wants our land. He wants our water. He wants his very own personal United States senator. He is trying to buy Texas."

northjersey.comSee famous NY Knicks 'Tip-in' at this Halloween house in NJNY Knicks tip-in victory as a Halloween decoration display at the corner of George Street and Westervelt Ave in Tenafly on Friday, Oct. 16, 2026.

The "Wemby" skeleton lol. www.northjersey.com/picture-gall...

nytimes.comTrump Endorses Netanyahu as Best U.S. Partner in Tight Israeli ElectionThe president praised the embattled prime minister for his endurance, saying any other candidate would be the “wrong Leader.”

By far not the only losing candidate he is endorsing this fall www.nytimes.com/2026/10/18/u...

Mor Naaman reposted

Cornell Tech

@cornelltech.bsky.social

Cornell Tech’s Master’s Programs Open House will take place on October 19 at 4:30 p.m. ET on Roosevelt Island. Details about the event and registration are available at gradprofessional.cornell.edu/register/?id...

I'm with Aram and openly disappointed with those of you -- some people I know and otherwise respect -- who aren't.

Aram Zucker-Scharff@chronotope.aramzs.xyz

I think some people are on there because their jobs told them they have to be and there's nothing that can be done about that. But to make it the center of my social media activity, or even an active account, would be an inherent betrayal of my moral core. It's that simple.

Happy that @nathumbehav.nature.com put out the research briefing describing our paper -- this is a one-pager open-access summary: rdcu.be/1X2W1SuAH25a

rdcu.berdcu.be

Mor Naaman@informor.bsky.social

🆕 📝 New paper officially out in @nathumbehav.nature.com! In 2023, Change.org built an AI-assisted writing tool into their platform. We offer causal evidence that access to the AI tool changed petition text (word count, language)... but may have actually hurt outcomes (signatures, comments).

share.googleAI coding agents generate more code, but not more softwareStudy finds coding efficiency gains get "absorbed" by human review "bottleneck."

Weirdly, same trend and similar takeaways (and methods!) in the research summarized here: AI coding agents generate more code, but not more software - Ars Technica share.google/tqH0TzAJJpeR...

Mor Naaman@informor.bsky.social

For more on this trend, see this paper by Cui et al. Here they show that AI assistance in writing cover letters helped with callbacks -- but only for the first couple of months. arxiv.org/abs/2509.25054

Bombshell lawsuit in Israel 2 weeks before the elections: the Oct 7 Shin Bet leader Ronen Bar filing a defamation case against Netanyahu, who has been spreading lies against the defense leaders trying to deflect ANY responsibility. Includes an extensive timeline of all warnings Netanyahu received.

The filing states he is doing it in lieu of an official State Commission of Inquiry (which Netanyahu refused to set up, naturally). Also, the lawsuit requests a comically low amount by US standards: $1M.

citizensandtech.orgIncident Reporting Beyond Frontier ModelsAs concerns grow about the risks of AI models, how can we notice when they go wrong and prevent catastrophes in the future?

I get "we used our models to understand the device and where it had come from" (though entirely irrelevant!) but "...identifying that a specific employee had planted it a few days earlier"? With "our models"? Wtf?

Carl Quintanilla@carlquintanilla.bsky.social

* OPENAI: SOPHISTICATED RECORDING DEVICE HIDDEN IN OUR OFFICES

Mor Naaman reposted

J. Nathan Matias

@natematias.bsky.social

As more efficient, capable AI models are given more power and autonomy in the form of agents, how can we notice when they go wrong and prevent catastrophes in the future? Baobao Zhang & I have written a new report on incident reporting for open models & agents. citizensandtech.org/2026/10/inci...

Mor Naaman reposted

Wedding in Philly. "Dearly beloved, we are gathered here today 9 days before the opening of the season 11 days before the Sixers' first home game To witness the joining in the holy matrimony of..." True story.

nature.comIntroducing AI to an online petition platform changed outputs but not outcomes - Nature Human BehaviourCorpus et al. examine how the introduction of an AI writing tool impacted Change.org petitions, finding that it increased petition homogeneity but did not improve petition outcomes.

Paper and materials here; if you don't have access, you can find the unofficial pre-print version on my website. www.nature.com/articles/s41... Happy to hear feedback and ideas!

Dang it here's a better link (without the ref to the paper section included in the previous one). www.nature.com/articles/s41... Also, my website now linked from my profile.

Isabel came up w/ another analysis, looking at repeat petition writers who returned to write a 2nd petition. We compared people who wrote 2 petitions before the AI launch w/ those who wrote 1 before and 1 after AI feature access, confirming our results: AI access changed the text, not the outcomes.

Why was the AI not helpful? See the paper for more discussion, but some possible reasons (we are looking at now) include decreased specificity of petitions and (I think, more significantly) reduced effort by the petition writers, who were less committed after AI wrote for them.

Was the AI assistance helpful? In general, petitions were NOT more successful with access to AI -- if anything, our results show that petitions may have been less successful in hitting signature and comment thresholds (see the end of this thread for speculation as to why this was the case).

One could ask whether these diminished outcomes are due to new authors joining the platform with the increased access that writing with AI enables (which is one of the reasons the feature was launched in the first place). We saw no evidence of a widespread increase in participation. Further...

As you can imagine, the petitions' textual/lexical aspects shifted with access to the AI feature: they were, on average, 50 words longer with access to "write with AI"; the sentences were longer, and the words were more complex. At the same time, the petitions became more similar to each other.

(In the supplementary materials, you can also see an increase, post-AI, in the use of words such as "implement", "urge", and "mandate" in the post-AI petitions' titles).

This was work led by the amazing @isabelcorpus.bsky.social with @allisonkoe.bsky.social and @eegilbert.org. Looking at the data we crawled (millions of petitions), Isabel noticed that the textual changes lagged in Australia; we later confirmed this was due to late launch of the AI feature there.