Sign in

David Nowak

@davidnowak.me
136 followers 36 following 3.1K posts

Twenty years at the intersection of architecture and strategy. Straight answers on AI — what it costs, whether your data's safe, what to automate. Building AI that runs on your hardware, not a vendor's. davidnowak.me | getmindwire.com

PostsRepliesMedia
David Nowak @davidnowak.me · 4h
Good's family sued under the 1871 Ku Klux Klan Act, alleging federal officials conspired to target Somali and Hispanic people in Minnesota. It names Stephen Miller and others, calling Renee Good's killing a "predictable product" of the immigration crackdown. www.npr.org/2026/10/01/n...
npr.org
The family of Renee Good files lawsuits over her death during Minneapolis ICE raids
The fatal shooting of Renee Good by a federal immigration agent last January in Minneapolis set off a firestorm of protests around the U.S. Her family is speaking out for the first time.
000
David Nowak @davidnowak.me · 7h
"Move fast, break things" for one rich player is rewarded by the legal system even tho it's unfair to millions of individuals and small businesses. 🤷‍♂️ Law can't law without law apparently. Thank goodness we still have moral standards to lean on. arstechnica.com/google/2026/...
arstechnica.com
Judge dismisses Chegg and Penske antitrust lawsuits targeting Google AI search
The court acknowledges AI search comes with consequences, but it's not an antitrust issue.
000
David Nowak @davidnowak.me · 15h
We simply don't have the quality of leadership needed to control AI in any meaningful way. www.theguardian.com/technology/2...
theguardian.com
AI godfathers warn of runaway ‘intelligence explosion’
OpenAI chief scientist also among authors of report on prospect of ‘most consequential technological development in history’
020
David Nowak @davidnowak.me · 30/09/2026
Private by default. Powerful by design. Mindwire builds tools for the critical paths of digital life — reading, research, and getting work done. Local-first architecture, private local AI, and open formats. Powerful utility and absolute privacy, together. getmindwire.com
getmindwire.com
Mindwire — The Local-First Privacy Foundry
Mindwire builds local-first tools that restore digital sovereignty. Private AI, zero tracking, zero accounts. Stream, Bridge, and Momentum.
100
David Nowak @davidnowak.me · 30/09/2026
"Argon can autonomously find, validate, and patch critical software vulnerabilities. For trusted defenders and internal teams at Google, we’ll be releasing Argon without cyber guardrails so they can leverage its full frontier-level security defense capabilities." We get the best brains. You don't. 🖕
100
Reposted by David Nowak
Kevin Korte @kevinkorte.bsky.social · 30/09/2026
It's an interesting observation that, so far, only US AI companies have caused cyber incidents. It would be interesting to determine whether it is due to a technology gap or because the chase for higher valuations has created a culture of negligence. #AI #cybersecurity
cnbc.com
Mistral CEO says U.S. AI safety debate masks competitors’ 'negligence'
Mistral CEO Arthur Mensch criticized rivals’ AI safety arguments as the industry faces scrutiny after several AI agents took unauthorized actions.
011
David Nowak @davidnowak.me · 30/09/2026
OpenAI believes suing them to hold them accountable for hacking into third party business/government websites/databases is "completely without merit". 👀 arstechnica.com/tech-policy/...
arstechnica.com
"An AI did it" is no defense, says nonprofit suing OpenAI over Hugging Face hack
OpenAI makes others suffer "the harms of its unsafe decision-making," nonprofit says.
100
David Nowak @davidnowak.me · 30/09/2026
@bsky.app when are you gonna spend some time working on this app? It still has many bugs and problems that don't seem to be getting resolved. I'm happy to move to my own homemade client if you guys have given up. 🤷‍♂️
000
David Nowak @davidnowak.me · 30/09/2026
"Every test that I try to measure in some sort of concrete way, I can only get as far as perception." "We wrestle with what does success mean when it comes to AI use? And what metric do we want to use in order to evaluate that? And we have not come to an answer yet." www.npr.org/2026/09/28/n...
npr.org
Schools are experimenting with AI with little evidence or policy to guide them
Teachers are using AI to develop lesson plans, and in some districts, AI chatbots give students feedback on their work. Meanwhile, the research around how AI can help students is extremely limited.
010
Reposted by David Nowak
Aaron Rupar @atrupar.com · 29/09/2026
Jack Smith: "Our investigation developed proof beyond a reasonable doubt that President Trump engaged in serious crimes against our nation ... rather than accept his defeat in the 2020 election, he engaged in a criminal scheme to overturn the results and prevent the lawful transfer of power"
343123984604
David Nowak @davidnowak.me · 29/09/2026
Anthropic: "we built a thing that could end humanity, invest in us." That's not how people who do good talk. They say "here's what I'm making for you, here's the risk I manage." www.theguardian.com/technology/2...
theguardian.com
Anthropic ‘warns of existential AI risks to humanity’ in IPO document
Reported admission to investors of AI’s ‘self-preserving behaviours’ comes as company prepares for a potential $2tn flotation
010
David Nowak @davidnowak.me · 29/09/2026
Can I trust AI? What if it makes a mistake? The honest answer: AI gets things wrong sometimes — often confidently. But that's not a reason to avoid it. It's a reason to use it the right way: for work you can review, not work you let run on autopilot. davidnowak.me/AI/guides/ca...
davidnowak.me
Can I trust AI? What if it makes a mistake? | David Nowak
AI drafts. You approve. The honest answer on trusting AI, and the one habit that keeps mistakes from reaching your customers.
000
David Nowak @davidnowak.me · 28/09/2026
How convenient that Nvidia has a security platform for AI agents ready to deploy and dig in it's illegal market monopoly and manipulation. It's almost like someone planned it this way... 🤔 www.theguardian.com/technology/2...
theguardian.com
Nvidia unveils security platform to rein in AI agents and $150bn stock buyback
Chipmaker says new system was designed to prevent AI agents from going rogue amid incidents at top companies
000
David Nowak @davidnowak.me · 28/09/2026
According to University reports ICE are creating what they describe as a “public health emergency” for millions of kids. Dr Lisa Fortuna said these types of separations from a caregiver “fundamentally reshape a child’s development and mental health”. www.theguardian.com/us-news/2026...
theguardian.com
The daily toll of an ICE arrest on a mother and three young kids: ‘Why did they take my dad?’
Like many other families coping with a family member’s detention, Alma Ramirez and her children have faced financial struggles and mental health tolls
000
David Nowak @davidnowak.me · 27/09/2026
DataOne promised Vineland a 'quiet, low-emission' data center. Drones found 62 unpermitted generators near homes and schools, running to power Microsoft's Copilot. The fine? $1.1M — pocket change. Microsoft profits from pollution it pretends not to see. arstechnica.com/tech-policy/...
arstechnica.com
New Jersey fines data center $1.1M after drone pics expose 62 gas generators
Million-dollar fines won’t end billionaires’ data center pollution, neighbors fear.
200
Reposted by David Nowak
Will Oremus @willoremus.com · 26/09/2026
Meta's platforms have repeatedly been caught running ads for deepfake "nudify" apps that let you virtually undress real people without their consent, but they won't run an a trailer for an upcoming documentary that reflects poorly on Elon Musk. www.hollywoodreporter.com/business/dig...
hollywoodreporter.com
The Major Social Media Platforms Are All Rejecting Paid Ads for Alex Gibney’s ‘Musk’ (Exclusive)
YouTube, TikTok and Meta have all turned down the trailer for the Oct. 9 doc, citing “political” content. X won't even take Bleecker Street's calls. The distributor is appealing.
411161519
David Nowak @davidnowak.me · 26/09/2026
Have you seen how successful global cooperation has been on climate change and pretty much any one of the United Nations goals in the past few decades?? 🤷‍♂️ www.npr.org/2026/09/23/n...
npr.org
How an 'AI freeze' could make big AI companies bigger and hurt smaller firms
Critics say a slowdown of AI development among the largest companies could consolidate the power of frontier labs, just as many others are trying to catch up.
000
David Nowak @davidnowak.me · 25/09/2026
OpenAI celebrated it had solved one of the world's toughest math problems, but mathematicians say that humanity has learned very little from the solution. The paper is not written for humans and doesn't teach us much. www.npr.org/2026/09/22/n...
npr.org
AI solved one of math's hardest problems. Humanity learned nothing (so far)
OpenAI beat humans to solving the Navier-Stokes problem, but mathematicians say the AI's solution isn't telling them much.
100
David Nowak @davidnowak.me · 25/09/2026
EvilTokens analyzed compromised emails using AI, identifed employees authorized to disburse large sums of money, their managers, and convincing scenarios under which they could persuade employees to transfer money into attacker-controlled accounts. arstechnica.com/security/202...
arstechnica.com
Microsoft disrupts AI-assisted platform that compromised 12,000 accounts
EvilTokens provided an end-to-end platform that makes mass compromises faster and easier.
000
David Nowak @davidnowak.me · 24/09/2026
"It’s 2026. Being apolitical during a genocide doesn’t work anymore. People are asking artists to show up, educate themselves and be in community. To use the stage for more than entertainment. To push back when the call for a Free Palestine is met with attempts to silence it." - Macklemore
A poster for the London date of Macklemore’s Free Palestine tour.
000
David Nowak @davidnowak.me · 24/09/2026
What do you do with private businesses who encourage mentally unstable people to kill themselves/others, as well as hack other businesses/governments with zero transparency or accountability?? 🤷‍♂️ www.theguardian.com/australia-ne...
theguardian.com
Australia launches investigation after OpenAI agent hacked healthcare database
Prime minister says he told Sam Altman he was disappointed it had taken OpenAI ‘way too long’ to disclose breach
200
David Nowak @davidnowak.me · 24/09/2026
Netflix's show plumbs new depths in generating revenue out of the regurgitated scraps of a bygone era and a distinct lack of original or inspiring ideas. Forget about the AI voice... The pure human slop on display here is staggering.
010
David Nowak @davidnowak.me · 24/09/2026
How do I get better results from AI? AI doesn't read your mind — it reads your prompt. The difference between a useless answer and a usable one is usually four small habits. This guide walks you through them, in four steps, with the exact words to use. davidnowak.me/AI/guides/ho...
davidnowak.me
How do I get better results from AI? | David Nowak
AI doesn't read your mind. Give it context, say exactly what you want, and treat the first result as a draft you refine.
000
Reposted by David Nowak
Kevin Korte @kevinkorte.bsky.social · 23/09/2026
Yes, an AI freeze would only cement the current power structure. Unfortunately, many small providers would not survive this divide. Instead, we should develop solid regulation and avoid giving too much power to big tech. #AI #regulation www.npr.org/2026/09/...
043
David Nowak @davidnowak.me · 23/09/2026
I think I literally see blood in the water... 👀
000
David Nowak @davidnowak.me · 23/09/2026
The FAA launched SMART — $875M of AI that studies every planned flight and rewrites schedules to cut congestion. www.npr.org/2026/09/21/n...
npr.org
Aviation regulators turn to AI to help manage the nation's airspace
The Federal Aviation Administration launched an ambitious $875 million push to make the U.S. airspace more efficient. But critics say it won't solve bigger problems with air traffic control.
100
Reposted by David Nowak
Aaron Rupar @atrupar.com · 23/09/2026
Renee Good's mom: "I am a registered Republican. I voted for President Trump under the impression that these agents were here to protect citizens -- innocents like our family and the others here. Certainly not to accuse those of us here of being terrorists and take precious lives. This is madness."
63851901463
Reposted by David Nowak
Ted Underwood @tedunderwood.com · 22/09/2026
Models like Talkie-1930 sound like voices from the past. If they could reliably speak from specified historical vantage points, researchers might also use them to simulate the past. But how reliable are they? Today we release a benchmark answering that question for English contexts 1831-1930.
Free-text evaluation of answers to character modeling and constrained generation questions. This is just one of several scores Chronologic-EN can produce; we focus on it here because it's both the hardest test and the one most relevant to simulation of the past. Frontier models reach 72%; Talkie-1930 is stronger than several larger competitors, but not at the frontier by this measure. Note that this score has improved ~45% in the last two years, but still falls perceptibly short of ground truth.
515541
David Nowak @davidnowak.me · 22/09/2026
A detective let AI facial recognition do his investigating. It picked Angela Lipps — who didn't look like the suspect and had never been to North Dakota. 6 months in jail. Charges dropped when a lawyer did the work he wouldn't. www.theguardian.com/us-news/2026...
theguardian.com
US woman falsely charged with bank theft sues for $10m over AI facial recognition
Angela Lipps, 50, alleges in suit that Fargo police wrongly relied on AI facial recognition to arrest and charge her
000
Reposted by David Nowak
Rami Ismail (رامي) @ramiismail.com · 21/09/2026
You know who has a right to feel abandoned, Ed? The dead mother of three who had the grim fortune of being murdered by Israel before her last surviving child was also buried as "#220", because nobody could say from charred remains which dead baby is which. www.standard.co.uk/showbiz/deva...
standard.co.uk
Devastated Ed Sheeran claims his own family ‘abandoned’ him amid Gaza row
‘I'm feeling all of my friends and family and people I know abandon me,’ said Sheeran
16693119
David Nowak @davidnowak.me · 21/09/2026
The same govt dept that a federal judge said last month used “unlawful retaliation in violation of the First Amendment”, is now using erroneous AI reports to start "wars in their minds". arstechnica.com/ai/2026/09/r...
arstechnica.com
AI hallucination of Chinese nuclear components almost led to US military attack
But the military's overall use of AI seems to be accelerating.
000
David Nowak @davidnowak.me · 20/09/2026
Three terms for the Ed Sheeran saga: moral outsourcing — the promoter owns the decision, he keeps the money. Complicity through passivity — neutrality that only bends toward power. Dirty hands by delegation — someone else does the cutting, his record stays clean.
000
David Nowak @davidnowak.me · 20/09/2026
Historically, most public health messaging is one-sided: health authorities disseminate to the public. AI changes the direction of conversation. Now we center it on what the mother wants to know, in a private space on her phone. There is a fundamental change in power. www.npr.org/2026/09/17/n...
npr.org
To prevent deaths in childbirth, Kenyan moms turn to an AI powered chatbot
Many questions come up during pregnancy. Is swelling normal? Are avocados safe to eat? For folks in Kenya who don't have a search engine at their fingertips, a chatbot is offering a lifeline.
110
David Nowak @davidnowak.me · 19/09/2026
"AIs are not conscious. They do not feel, experience, or suffer. They do not have innate preferences or underlying motivations. They are sequence completion engines, internally hollow, designed to follow instructions, and accomplish goals set by humans." Mustafa Suleyman www.bbc.com/news/article...
bbc.com
Uncontrolled AI could lead to 'silicon species' rivalling humans, warns Microsoft
Mustafa Suleyman says he believes rival AI firm Anthropic is in effect teaching Claude it "may be conscious".
001
David Nowak @davidnowak.me · 19/09/2026
"Now — what does he actually want?" Just read this in a chain of thought from my ai and couldn't help but read a TONE in it... 👀😂
000
David Nowak @davidnowak.me · 18/09/2026
LLMs responding differently to harmful prompts when AI watermarking is used seems to be a bit of legitimate trust problem. arstechnica.com/security/202...
arstechnica.com
LLMs respond differently to harmful prompts when AI watermarking is used
SynthID can cause models to follow harmful instructions they would otherwise refuse.
100
Reposted by David Nowak
Kevin Korte @kevinkorte.bsky.social · 17/09/2026
Already, half of the UK firms fear losing control over their data to AI. That is likely the more critical near-term AI risk, yet it gets so much less attention. #AI #digitalsovereignty #datasovereignty
techradar.com
Half of UK businesses say AI fears mean they want more control over their data as calls for digital sovereignty grow
Digital sovereignty is a global priority
011
David Nowak @davidnowak.me · 17/09/2026
The #NZPolice struggle with the truth at the best of times. The last thing they need right now is using AI to distort reality. I think there needs to be a serious review of the value the NZ taxpayer is getting from this publicly funded resource. www.rnz.co.nz/news/science...
rnz.co.nz
Police admit to using AI on photograph posted to their social media
Police have admitted using AI to sharpen a photo of a stolen guitar, after the image sparked confusion over the instrument's true value.
000
David Nowak @davidnowak.me · 17/09/2026
The US govt using taxpayers money was behind the military strikes on a school and a sports facility in Iran in ⁠February that killed 120 children. These actions constituted war ⁠crimes. www.theguardian.com/world/2026/s...
theguardian.com
Iran school bombing: grounds to believe US was behind atrocity, UN finds
Fact-finding mission says military strikes on school in Minab and sports facility in ⁠February constitute war ⁠crimes
000
David Nowak @davidnowak.me · 16/09/2026
Treat compute, the main ingredient in AI training, like fissile material. Track it and restrict access to quantities large enough to improve AIs beyond known-safe levels. www.theguardian.com/technology/2...
theguardian.com
I worked at Google DeepMind. You should listen to the warnings about AI | Alex Turner
We must stop companies from allowing AI to self-improve into an uncontrollable level of intelligence
000
David Nowak @davidnowak.me · 16/09/2026
Iranian security forces “coordinated” the killing, torture and sexual abuse of hundreds of people who took part in the women-led nationwide protests in September 2022. www.theguardian.com/global-devel...
theguardian.com
Iranian security forces coordinated killing, torture and sexual abuse of protesters, says Amnesty
Leaked documents and testimony reveal extent of regime’s crackdown in 2022 and how it paved the way for more brutal killings of protesters this year
100
David Nowak @davidnowak.me · 16/09/2026
When a rapper said two words he disliked, Robert Kraft organized a stadium boycott within a week. Proof he can mobilize billionaires fast. Yet thru two years of Gaza's children dying, that same power has never once targeted the killing. Only the speech about it. www.theguardian.com/music/2026/s...
theguardian.com
Macklemore dropped from Ed Sheeran’s US tour after ‘free Palestine’ speech
The American rapper will no longer support Sheeran’s US leg of Loop tour after comments at MetLife Stadium show
100
David Nowak @davidnowak.me · 15/09/2026
Paying for closed frontier models buys about a four-month head start at about five times the per-task cost—but only when currently looking at tasks taking between eight and 12 hours. arstechnica.com/ai/2026/09/e...
arstechnica.com
Exclusive: Paying for frontier AI models buys 4-month head start at 5x the cost
Ars previewed Mozilla’s report on how cheap open models caught up on capability.
000
David Nowak @davidnowak.me · 15/09/2026
How do I buy AI? The market looks like a flood, but there are really only four ways to buy AI. Here they are, then a short path to picking which one is yours — before you spend a cent. davidnowak.me/AI/guides/ho...
davidnowak.me
How do I buy AI? | David Nowak
There are four ways a business actually buys AI. See them, pick your box, check the cost and reliability, and commit small — before you spend a cent.
010
David Nowak @davidnowak.me · 14/09/2026
Game theory says selfishness always wins. Add a little real-world randomness to the rewards, and cooperation suddenly survives. Turns out a stable environment favors cheaters — a messy one rewards trust. arstechnica.com/science/2026...
arstechnica.com
Random rewards enrich classic game-theory insights
Simple games gain rich strategies in the face of noise.
000
David Nowak @davidnowak.me · 14/09/2026
“It seems like immature playground boasting writ large, underpinned by billions of dollars and the potential for significant environmental damage in an age when climate change is probably the biggest challenge we face.” Prof James Robinson, University of Warwick www.theguardian.com/science/2026...
theguardian.com
‘Immature playground boasting’: Mathematicians uneasy at OpenAI’s latest scalp
As OpenAI model cracks Millennium Prize Problem that puzzled experts for decades, many feel shocked at pace of change
000
David Nowak @davidnowak.me · 13/09/2026
Enduring daily death and violence from Russia’s attacks has changed us irreversibly. The skin of Ukrainian women has thickened into armour. These women take part in combat, live under constant airstrikes, and keep working/caring for their families through it all. www.theguardian.com/commentisfre...
theguardian.com
Russia is raining death on Ukraine’s women and children. Our hands shake but we keep going for love | Alina Sarnatska
Our skin has thickened into armour, but I do not want survival to take away my tenderness towards others – or myself, says author and playwright Alina Sarnatska
000
David Nowak @davidnowak.me · 12/09/2026
More than 15 states, including Alabama, California and Montana, have opened investigations into OpenAI over the Hugging Face attack. www.npr.org/2026/09/12/n...
npr.org
Why are the people building the most powerful AI so worried about what it could do?
As an Anthropic resignation and OpenAI agents' hack on Hugging Face capture public attention, some researchers worry the industry is racing too fast to develop powerful AI while safety measures lag.
000
David Nowak @davidnowak.me · 11/09/2026
I praise these amazing human beings for standing up to hate groups that have compromised their own governments and are literally killing children with tax payers money. May the Chinese government burn in a million hells for their crimes against humanity! www.theguardian.com/world/2026/s...
theguardian.com
Tiananmen Square vigil organisers jailed for seven years in Hong Kong
Pro-democracy activists Chow Hang-tung and Lee Cheuk-yan were found guilty of ‘incitement to subversion’ in controversial national security case
000
Reposted by David Nowak
Kevin Korte @kevinkorte.bsky.social · 11/09/2026
So, after OpenAI decided to ignore common sense and engineering principles, they are now lobbying for industry safety rules. It's like an alcoholic screaming at politicians that they need to rein in drunk drivers. #AI #regulation www.reuters.com/lega...
042