Nirit Weiss-Blatt, PhD

Here are 10 additional media studies on generative AI coverage: aipanic.news/p/what-10-stud…

@random_walker A very thoughtful presentation. Thank you for sharing it.

I had the honor of giving a keynote at the International Conference on Machine Learning in Seoul last week titled “What will be left for us to work on?” I addressed the widespread anxiety about how we should adapt as AI capabilities increase. I was thrilled by the talk’s reception, so I have made my slides available, annotated with a lightly edited transcript: cs.princeton.edu/~arvindn/talks… I made three arguments. First, the "AI as Normal Technology" framework is a correct and useful as a way to think about AI’s impacts, unless and until there is some future discontinuity such as through recursive self-improvement. Second, even though we should take recursive self-improvement seriously, there is no milestone that companies might achieve in the lab that will suddenly put us all out of work. Third and finally, jobs of the future will be radically different, and a lot of adaptation will be needed. I shared my thinking about what this might look like and ended with a vision of human/AI “co-superintelligence”.

@sharongoldman Spot on 🎯

"I would rather trade places with my grandkids than with my grandparents." Prof. Sheizaf Rafaeli wrote this in the Pew Research Center's report "The Future of Well-Being in a Tech-Saturated World" in 2018. I keep repeating it because this one simple sentence captures an entire point of view, grounded in historical knowledge. Don't get persuaded by doomers who don't want to have kids. Bet on a future of opportunities, and (if you can) help bring life to this planet. It's the most impactful (and magical) thing to do.

@Josh_Koehn x.com/drtechlash/sta…

NEW: Dustin Moskovitz, billionaire Facebook co-founder, likes to eat meat. But the Democratic megadonor has poured $480M into activist campaigns that contribute to rising cost of eggs & meat — while his foundation holds $89M stake in Impossible Foods nypost.com/2026/07/11/us-…

3. Manufacturing the appearance of a broad grassroots movement. As quoted in the New York Post: "That image badly understates the sophistication and funding behind the movement." It's the same well-oiled machine in the AI existential risk movement: An interconnected ecosystem of nonprofits, think tanks, university centers, fellowships, communications groups, and advocacy campaigns financed by the same billionaire. nypost.com/2026/07/11/us-…

This story is about how Dustin Moskovitz spent nearly half a billion dollars on animal-welfare activism, which critics say has driven up costs and hurt everyday Americans. But replace animal welfare with AI x-risk and Impossible Foods with Anthropic, and it's the same M.O. 1. Fund campaigns for tighter AI regulation while being deeply invested in the company positioned to benefit from it. 2. Fund the information ecosystem – journalism, communications organizations, and policy institutes – in order to shape the messaging. The issue is framed as an overwhelming moral emergency, making extraordinary restrictions seem politically necessary.

@tszzl Why are you interested in Kabbalistic Creation?… As a Jewish woman, I can clarify that * It is supposed to be אין סוף/ein sof and not אין סוך/ein soch (probably a typo). * The meaning of אין סוף/ein sof is infinite. * And here are the meanings of the rest of the Hebrew words:

Lightcone Infrastructure

@WSJ Great coverage of Sam Kirchner's story. You can read more about it here: techdirt.com/2025/12/05/rad…

A mysterious disappearance hangs over the activists fighting to stop AI on.wsj.com/4bgSTxC

@zephyr_z9 The ramifications of Anthropic's IPO could be crazier (and bring unprecedented power to already power-seeking doomers)

@aaronscher @cogsciclinton @taig_icml Aaron, “the case for those assumptions” was unconvincing (to many). aipanic.news/p/why-the-doom…

@cogsciclinton @DrTechlash Sure, if you disagree with the assumptions then it makes sense you don't find the paper useful! (Fortunately for me, the peer reviewers for @taig_icml thought the work was useful). If you want to hear the case for those assumptions, I recommend this book ifanyonebuildsit.com

The Machine Intelligence Research Institute calls for restricting AI research itself. It lays out a research-control regime for monitoring researchers and organizations, including penalties that "could plausibly include prison sentences." This new paper catalogs 28 mechanisms, including intelligence gathering, international search warrants and inspections (of properties, computers, and files), polygraphs; inference-content monitoring (of user prompts, tool use, model outputs), sting operations, AI-assisted code review (which may extend to codebases, experiment logs, and research documents), embedded auditors, employee-monitoring software, chip-use monitoring, training-data review, and model allowlists.

@lifeext Yeah, it’s explicit in MIRI’s (2024) communications strategy: “Many other organizations are attempting the coalition-building, horse-trading, pragmatic approach… We don’t play 4-D chess. We don’t conceal our true intentions.”

Well, at least they are frank about what they are calling for

MIRI's blueprint for an AI surveillance regime The common theme across the new 6 papers is MIRI's effort to normalize extreme surveillance under the banner of verification. The Orwellian proposals are all over the stack: compute layer (track chips and monitor clusters); workload layer (detect ML training using telemetry); network layer (tap and hash all data center I/O); inference layer (reproduce AI outputs); and research layer (restrict AI research using intelligence tools, search warrants, and code review). It's another example of AI "safety" being used to justify AI authoritarianism. The answer to speculative machine tyranny should not be actual human tyranny. We need to defend open research, the right to compute, market competition, and broad access.

MIRI (the doomer organization of Eliezer Yudkowsky & Nate Soares) published new papers in which its "Technical Governance Team" discusses the following restrictions and bans: 1. The "Closing Window" paper proposes frontier training bans, forced hardware consolidation into monitored data centers, research restrictions, inference monitoring, and even physical disconnection of AI chips. "Research that would undermine this regime, such as research increasing general-purpose AI capabilities or advancing distributed training, is restricted." 2. The "Distributed Training" paper proposes registration of smaller clusters, location verification, and short-notice inspections. It quotes MIRI's previous recommendations for banning "model pre-training over 10²⁴ FLOP and fine-tuning over 10²³ FLOP," monitoring "model training between 10²² FLOP and 10²⁴ FLOP," and registering and monitoring "all clusters of chips with more computing power than the equivalent of 16 H100 GPUs." 3. The "Bit-Exact AI Inference" paper proposes a regime in which auditors can verify exactly what computation happened on AI hardware: A forensic AI compute audit in which inference claims become auditable and pass/fail verifiable, including through logs of the hardware/software stack, model weights, topology, versions, and batch size. 4. The "Detecting Hidden ML Training" paper develops a classifier and recommends tamper-resistant GPU telemetry with authenticated reporting channels and secure-booted monitoring software. 5. The "Fingerprinting All AI Clusters" paper adds network-level data-center surveillance. It proposes to place network taps on all information-carrying wires between the AI cluster and the outside world.

Interesting to revisit AI Panic's old posts: ▪️Oct 2023: The frontier lab's fearmongering ▪️April 2024: The doomers' lobbying ▪️June 2024: Yoshua Bengio's testimony asks to limit people's access to powerful AI ▪️April 2025: The doomers' shift to NatSec (to justify previous authoritarian proposal)

Lessons from the current situation: 1. The predictable outcome of AI doomerism was always gatekeeping. 2. AI panic has become the business model of frontier labs. 3. This power grab should be fought through decentralization: Open-source AI, real competition, and broad access.

So saddened by the news about Om Malik (@om). I have followed his work for two decades, since the early days of the tech blogosphere. We all learned so much from him. I admire his writings, his unique voice as an opinion leader, and his thoughtfulness. He was such a legend. RIP. om.co/2026/06/24/196…

This is an amazing new detail about the Zizians member who has now officially been charged with killing her parents: There was a list on Michelle Zajko's phone titled "There are so many things we f-- up" that "detailed missteps, including not taking shell casings from the homicide scene."

@zeldapoem FYI @PessimistsArc @LouisAnslow (This cool website was made by Prof. Andrew Maynard from Arizona State University)

Just found a gem of a website "Technology-Driven Moral Panics" or how societies all the way back to Ancient Greece worried about new inventions dooming humanity