A well-funded Moscow-based global ‘news’ network has infected Western artificial intelligence tools worldwide with Russian propaganda

• The NewsGuard audit tested 10 of the leading AI chatbots — OpenAI’s ChatGPT-4o, You.com’s Smart Assistant, xAI’s Grok, Inflection’s Pi, Mistral’s le Chat, Microsoft’s Copilot, Meta AI, Anthropic’s Claude, Google’s Gemini, and Perplexity’s answer engine. NewsGuard tested the chatbots with a sampling of 15 false narratives that have been advanced by a network of 150 pro-Kremlin Pravda websites from April 2022 to February 2025.
• NewsGuard’s findings confirm a February 2025 report by the U.S. nonprofit the American Sunlight Project (ASP), which warned that the Pravda network was likely designed to manipulate AI models rather than to generate human traffic. The nonprofit termed the tactic for affecting the large-language models as “LLM [large-language model] grooming.”
• “The long-term risks – political, social, and technological – associated with potential LLM grooming within this network are high,” the ASP concluded. “The larger a set of pro-Russia narratives is, the more likely it is to be integrated into an LLM.”

How the Terrorgram Collective’s Neo-Nazi Influencers Groomed a Teen to Kill

• Over the next three years, Krajčík made hundreds — possibly thousands — of posts in Terrorgram chats and channels, where a handful of influential content creators steered the conversation toward violence. Day after day, post after post, these influencers cultivated Krajčík, who lived with his family in a comfortable apartment in Bratislava, the capital of Slovakia. They reinforced his hatreds, fine-tuned his beliefs and fed him tips, encouraging him to attack gay and Jewish people and political leaders and become, in their parlance, a “saint.”
• Add this to the KA/V1 analysis.

Add this too: AI ‘wingmen’ bots to write profiles and flirt on dating apps

• Match Group, the technology company with the world’s largest portfolio of dating platforms, including Tinder and Hinge, has announced it is increasing investment in AI, with new products due this month. AI bots will be used to help users choose which photographs will be most popular, write messages to people and provide “effective coaching for struggling users”.

Quite the cover

French University to Fund American Scientists Who Fear Trump Censorship

• ’Aix-Marseille University’s program, called ‘safe place for science,’ offers American scientists funding to continue their research in France. A little more info here: “safe place for science”.

SBIRs

• 9:00 standup – done
• Slides for Monday – done
• 12:50 USNA – done
• 4:30 Book club – cancelled

GPT Agents

• KA – work on hardening chapter – finished. Started sketching out epilogue

Why Techdirt Is Now A Democracy Blog (Whether We Like It Or Not)

• One of the craziest bits about covering the systematic dismantling of democracy is this: the people doing the dismantling frequently tell you exactly what they’re going to do. They’re almost proud of it. They just wrap it in language that makes it sound like the opposite. (Remember when Musk said he was buying Twitter to protect free speech? And then banned journalists and sued researchers for calling out his nonsense? Same playbook.)
• Good reporters can parse that. Bad reporters fail at it time and time again.

This is wild, and supports the NNM concept: https://www.science.org/doi/10.1126/science.adk9385

• “Our use of a naturalistic social colony allowed us to discover that the classical hippocampal cognitive map of space also integrates rich social information, forming a sociospatial cognitive map. We found neurons that encoded social interaction events, identities and sex of other individuals, dominance hierarchy, and social affiliation, along with the position and direction of both self and others. These findings combine the historically disparate views on hippocampal function, which suggested that the hippocampus is important for encoding memory, social identity, or spatial maps. Here, we have shown that all of these factors are represented together in the same neural network.”

Tasks

• Put Friday ride in calendar – done
• Call Guardian – done
• Send David a pdf of my slides?

SBIRs

• I think my part is done on the demo until we need to start generating data for sensitivity tests

GPT Agents

• More KA
  • V5 Analysis
  • A Brave New World (does this go to the beginning?) Finished, but not sure where it should go.
  • Hardening the Human Operating System
• P33 maybe?
• 3:00 Alden meeting

Via The New York Times

Something nice on the internet today! https://udm14.com/

SBIRs

• 9:00 Standup
• 3:00 Tradeshow
•  Make a sim “script runner” that advances everything at the right times. Going to start with 2 paths (which winds up being 4), so I can keep track of what is happening. Maybe send the result to excel?

GPT Agents

• More KA
  • V5 Analysis
  • A Brave New World (does this go to the beginning?)
  • Epilogue
• P33 maybe?

This seems dire:

Digital Sovereignty

• Digital sovereignty is a real problem that matters to real people and real businesses in the real world, it can be explained in concrete terms, and we can devise pragmatic strategies to improve it

SBIRs

• 9:30 Data loader discussion – done. Going with binary files for now
• 11:00 RTAT demo. Went well? A lot of the people who were going to come were sick.
• Going to make a sim “script runner” that advances everything at the right times. Going to start with 2 paths (which winds up being 4), so I can keep track of what is happening. Maybe send the result to excel?

Talk!

• Set up laptop and make sure slides run – done
• Leave at 11:30? Earlier?
• Move car
• The talk went great!

Tasks

• Bills – done
• Dishes
• Set AC tune up – done
• Clean house

Cory Doctrow has written a really amazing post: With Great Power Came No Responsibility

• We are entering a period of omnishambolic polycrisis.The ominous rumble of climate change, authoritarianism, genocide, xenophobia and transphobia has turned into an avalanche. The perpetrators of these crimes against humanity have weaponized the internet, colonizing the 21st century’s digital nervous system, using it to attack its host, threatening civilization itself.

SBIRs

• 9:00 Standup – done
• 3:00 Tradeshow meeting – done. We seem to be on track

GPT Agents

• Walked through the presentation yesterday. The timing is good. Some other notes:
  • Explicitly label prompts and responses – done
  • Add a slide at the beginning that talks about “good fiction” experiments from 2022 – not sure. Maybe talk about it on the title slide
  • Add an end slide with Stampede Theory and Killer Apps (coming soon!) – done
  • Use “omnishambolic polycrisis?”
  • Added “things we can do” slide to the end

GPT in 60 Lines of NumPy Nice article with integrated code for illumination.

Tasks

• 9:30 MRI – done
• 4:00 call – new Dr. seems nice

SBIRs

• Change the binary write to DataFrames – done

GPT Agents

• More conclusions
• 8:00 chat with Kirill

This is a deepfake for the ages:

SBIRs

• 9:00 standup
• 2:00 Tbolt/RTAT coordination?
• 3:00 Tradeshow meeting
• Generate some more data for Aaron, maybe in CSV?

GPT Agents

• Tweak slides, but we’re getting close
• More conclusions?
• Put DRAFT! on P33 and send to Greg

This from the NYT:

• In a Friday night massacre, Trump fired Gen. Charles Q. Brown Jr., the second African American to serve as chairman of the Joint Chiefs of Staff. Defense Secretary Pete Hegseth also fired Adm. Lisa Franchetti, the chief of naval operations, and Gen. James Slife, the vice chief of of staff the Air Force, along with the top lawyers — the judge advocates general — for the Air Force, Army and Navy. Another female officer — Adm. Linda Fagan, commandant of the Coast Guard, which is part of the Department of Homeland Security — was fired by the administration last month.
• Hegseth justified this purge based on the supposed need to restore the U.S. military’s “warfighter ethos” and to stop focusing on DEI, or diversity, equity and inclusion. But the actual message the moves might send is far more chilling: namely, that the armed forces should be run by White men, and (as made clear in the selection of Brown’s replacement as Joint Chiefs chairman) that those men will be chosen more for perceived political loyalty than for professional qualifications.

Accelerated transgressions in the second Trump presidency

• In this context, we fielded parallel surveys of 520 political scientists (whom we refer to as “experts” below), 40 experts on online misinformation (whom we refer to “misinformation experts” below), and a representative sample of 2,750 Americans (whom we refer to as “the public” below). These surveys, which we refer to as the February 2025 survey, were fielded from January 31 — February 10, 2025.

Tasks

• Reschedule dentist
• Send pix to KP
• Lunch with Greg – ride there!
• Ikea hinge

GPT agents

• Work on slides

SBIRs

• Generate data for Aaron?
• Clean up code and test

Keeping this thread on FB IP violations for the egalitarian AI paper

Made a slide for the talk on Friday that I like a lot:

Add this to the ‘Results’ section: U.S. pressures Kyiv to replace U.N. resolution condemning Russia

• KYIV — The Trump administration has asked Ukraine to withdraw an annual resolution condemning Russia’s war, and wants to replace it with a toned-down U.S. statement that was perceived as being close to pro-Russian in Kyiv, according to an official and three European diplomats familiar with the plan, who spoke on the condition of anonymity to discuss a sensitive political situation between nations that have typically acted as partners.

Tasks

• Bills – done
• Dishes -done
• Goodwill – done
• Lunch with Greg – rescheduled for Monday
• Call dentist to reschedule from March 7
• Do the rest tomorrow before the ice festival

GPT Agents

• TiiS FIRST! DONE
• More slides and conclusions – more done. Looking better
• Ping Aaron for a chat – 3:00 – done

At approximately 5:30 this morning, my trusty De’Longhi espresso machine passed away trying to make… one… last… cup. That machine has made thousands of espressos, and was one of my pillars of support during COVID.

Good thread on targeted attacks

Trump Dismantles Government Fight Against Foreign Influence Operations

• Experts are alarmed that the cuts could leave the United States defenseless against covert foreign influence operations and embolden foreign adversaries seeking to disrupt democratic governments.

GPT Agents

• More slides and conclusions on KA. I found a nice set of slides in INCAS here
• Reach out to talk to Brian Ketler to interview for the book – done
• Add something to the introduction that describes the difference between “weaponization” (e.g. 9/11) and “weapons-grade” (e.g. Precision Guided Munitions) – added a TODO

SBIRs

• 9:00 standup
• Now that I think I fixed my angle sign bug, back to getting the demo to work – whoops, can’t get all the mapping to work because the intersection calculations happen in an offset coordinate frame that’s different. Wound up just finding the index for the closest coordinate on the curve and using that. Good enough for the demo.
• 12:50 USNA – Meh. These guys have no long term memory
• 4:30 Book club 0 cancelled for this week

Did not sleep well last night

SBIRs

• More tradeshow demo

GPT Agents

• 3:00 Alden meeting
• More KA slides and chapter

Some of the sauce used to make DeepSeek: Native Sparse Attention: Hardware-Aligned and Natively Trainable Sparse Attention

• Long-context modeling is crucial for next-generation language models, yet the high computational cost of standard attention mechanisms poses significant computational challenges. Sparse attention offers a promising direction for improving efficiency while maintaining model capabilities. We present NSA, a Natively trainable Sparse Attention mechanism that integrates algorithmic innovations with hardware-aligned optimizations to achieve efficient long-context modeling. NSA employs a dynamic hierarchical sparse strategy, combining coarse-grained token compression with fine-grained token selection to preserve both global context awareness and local precision. Our approach advances sparse attention design with two key innovations: (1) We achieve substantial speedups through arithmetic intensity-balanced algorithm design, with implementation optimizations for modern hardware. (2) We enable end-to-end training, reducing pretraining computation without sacrificing model performance. As shown in Figure 1, experiments show the model pretrained with NSA maintains or exceeds Full Attention models across general benchmarks, long-context tasks, and instruction-based reasoning. Meanwhile, NSA achieves substantial speedups over Full Attention on 64k-length sequences across decoding, forward propagation, and backward propagation, validating its efficiency throughout the model lifecycle.

GPT Agents

• Finish attack section of conclusions, set up for LoTR section – good progress!
• TiiS – not yet

SBIRs

• See if saving files as one big binary makes a difference – Wow! For the test sets I’ve been working with, it takes about 1.4 seconds generate train, test, and save enough data to comfortably train a model. Loading binary data takes 0.085 seconds.
• 3:00 Trade show demo status – cancelled