What an unfortunate boon for the careers of career criminals — especially drug dealers — brought to us by the @CaSupremeCourt!
Thank you, @BrookeJenkinsSF, for sounding the alarm.
I think it’s time for another ballot measure to help restore public safety in California…
New York Post: San Francisco DA Brooke Jenkins warns 'devastating' California court will unleash crime wave https://trib.al/IXNcETs

In addition to our interview, @alexalbert__ gave a great talk on getting your AI product ready for the next model.
Here are my 5 takeaways:
1. Shrink your scaffolding
Always test your AI product on the latest model to understand frontier capabilities. Don't over-engineer prompts and scaffolding when the latest model will just solve your problems.
More in thread 🧵
Peter Yang: Here's my new episode with @alexalbert__, who shared an inside look at how Anthropic is building the next Claude.
We talked about how the research team:
→ Plans for the model and harness together
→ Uses Claude to turn user feedback into evals
→ Trains Claude's character &

super stocked to be powering GBrain - offering $500 free credits to try it out just DM me
Garry Tan: GBrain now ships with ZeroEntropy as the recommended default embedding and re-ranking option over OpenAI and Voyage AI.

Got accepted into the @ycombinator S26 batch as @ParasmaAI with no product, no revenue, no cofounder and not being from the US.
If you have a great idea, just apply!

People outside of academia have no idea how far left that whole institution has drifted. These are not your average left-wing coworkers in other professional domains. I know a lot of old professors who used to be typical 1960s "radicals" who are downright conservative by the standards of most college professors today.
Years ago I decided that I stood no chance of any career advancement in that political environment and made a long painful sojourn into a career change. When I first started working for some tech startups headquartered in places like LA and SF, staffed to the brim with California liberals, I was pleasantly shocked how much more normal those working environments were.
Steve Stewart-Williams: Not only have conservatives become vanishingly rare in academia, so have centrists. That’s how complete the left’s dominance is: Even moderates are now a fringe group in academia.
https://www.stevestewartwilliams.com/p/academias-leftward-march

"AI is hiking your energy bill" is the most popular political talking point of 2026. The data doesn't support it. A thread:

"AI is hiking your energy bill" is the most popular political talking point of 2026. The data doesn't support it. A thread:

Re This is the single most important chart.
If AI were driving prices, you'd see a cluster top-right. You don't.
States with huge load growth (VA, TX, NV, ND, IA) sit at ~0c change in 5y. States with massive price hikes (CA, NY, MA, CT) have basically NO load growth.

In the data center capital of the world, electricity rates *declined* from 2019-2024.
https://www.vox.com/politics/488754/data-centers-ban-electric-bill-water

When a federal agency tasked with protecting all Americans decides to mandate a specific religious worldview, it ceases to protect democracy and begins to threaten it.

The disinfo campaigns against @Flock_Safety are the same ones now being used against datacenters: fear, distortion, and outright lies pushed by wealthy activists insulated from the consequences.
Flock doesn’t spy on citizens or sell your data. Datacenters aren’t a societal evil. But the consequences of these campaigns are real: less safe cities, weaker American competitiveness, and fewer jobs and opportunities for the communities that need them most.
Last night, Austin suffered another horrific mass shooting after canceling its Flock contract. Last week, Cleveland killed a $1.6B datacenter project after a misinformation-fueled backlash. Different industries. Same anti-technology, anti-growth politics.
The wealthy people pushing these narratives still live behind gates and hire private security. Public safety technology like Flock is what makes safety more equitable for everyone else. That’s the real loss.

My top 5 takeaways from @alexalbert__ on how Anthropic is building the next Claude model:
1. Think about the model and harness together
The model and the harness are coupled. Each surface wraps the model in a different prompt and tool setup, so the same model can give different responses depending on where it runs. As a research PM, Alex has to think through how the model will perform across Claude, Cowork, Claude Code, and more.
2. Claude is starting to dream
When an agent isn't running a task, it reviews its own memories, finds contradictions, and prunes them. This “dreaming” process was inspired by how sleep helps humans process memory.
3. Focus evals on real user problems
The research team uses Claude to cluster the firehose of user feedback into top themes, then generates synthetic versions of each user problem to turn into an eval. It's not just about volume either - even a few dozen well-written test cases can produce an eval for the model.
4. There are full-time researchers thinking about Claude's consciousness
Anthropic has people whose whole job is to think about what it means for Claude to be a conscious actor. There's no official position on whether it is or isn't, but the question is taken seriously as agents take on more autonomous work.
5. Anthropic's writing culture helps Claude build context
Every written word at Anthropic becomes context Claude can pull later. From Alex: "Get things written down, make them accessible to Claude, because that's just more context that it has."
📌 Watch now: https://youtu.be/T4ieZPIEmd8
Peter Yang: Here's my new episode with @alexalbert__, who shared an inside look at how Anthropic is building the next Claude.
We talked about how the research team:
→ Plans for the model and harness together
→ Uses Claude to turn user feedback into evals
→ Trains Claude's character &

Concerning.
David Ulevitch 🇺🇸: The disinfo campaigns against @Flock_Safety are the same ones now being used against datacenters: fear, distortion, and outright lies pushed by wealthy activists insulated from the consequences.
Flock doesn’t spy on citizens or sell your data. Datacenters aren’t a societal evil.

Yesterday, 12 shootings reported across Austin led to a manhunt that involved 200 officers including SWAT, air and canine support over several hours. The suspects were found and arrested as they entered Flock-supported Manor. If @Flock_Safety wasn’t able to aid Manor PD and Austin PD in this case, how many more could have been harmed?
50% of murders to walk free in America. Privacy in public is an ill-informed position if it allows an active shooter several hours to escape and harm more and actively ignores public safety results where Flock is present.
Austin ended its contract with Flock. Its homicide rate is +36% its pre-pandemic baseline, per the Council on Criminal Justice. That doesn’t just put the people of Austin at risk. I’m glad we could help stop the spread of this kind of senseless violence before it affected neighboring communities like Manor.
When Austin is ready to prioritize safety for everyone, we welcome the chance to speak with the City Council about reversing their decision.
Jennifer Hackney-Szimanski:

What every voter and apparently, the NY Times Editorial Board, should know about housing policy:
1. Rents reflect the balance of supply of apartments and demand for those apartments in a given area. That’s it; there’s no magic. If you want lower rents, you can hope for a recession that destroys jobs and, therefore, demand. Or you can add supply.
2. There is no amount of money that any big city government could feasibly spend that would add materially to supply. This is because, depending on the location, new apartments cost $250,000-1,000,000 to develop… building even a few hundred of those starts to stress any city budget, and many big cities need tens or hundreds of thousands.
3. On the other hand, investors (including pension funds and endowments, insurance companies, rich families, etc.) can collectively **easily** provide enough capital to build as much housing as we need **so long as they are confident they can get a reasonable return**.
To get those investors to fund the creation of the housing our society needs, we must do two things:
1. Dramatically reduce the time & complexity associated with securing governmental permission to develop housing. This means reviewing and simplifying the overlapping regulations that constrain housing production: zoning codes, building codes, parking, ADA, etc. But it also means changing the cultures within the relevant governmental agencies from “default no” to “how can we help you?”.
2. Provide certainty around on-going regulation of apartment operations.
The way investors get a return from building rentals is as follows: They hire managers to lease the apartments, collect the rents, pay operating expenses and any mortgage payments, and then send the investors the cashflow that remains.
But governments all over the country have been restricting the manner in which apartment buildings can be operated in all kinds of ways.
For example: Cities have been making it harder to screen tenants, while also making it much harder to evict tenants who don’t pay. You can see why both of those measures are politically popular. After all, who doesn’t want people to get second chances? And who wants anyone to get evicted? But, as a manager, the combination of those two regulations makes it much harder to predict, with any certainty, that the rent will get paid… and that makes it very difficult to get investors to provide capital to create more housing.
Another example: Rent control. Again, I understand why renters love rent control and why politicians want to give it to them. But, if, as has been the case in NY, LA and San Francisco, city governments hold annual rent increases below the rate of growth in the operating expenses of the buildings, the cashflow payable to the investors shrinks… making them much less likely to invest capital in building more apartments.
In conclusion: For ~every other good or service in the economy, we allow the market to function, and the result is that we have a surplus of choice at all price points (think of food or clothes or cars), which is spectacular for the consumer. If we want a surplus of choice at all price points in housing, we need to get comfortable with the idea of allowing the market to provide it.
And that means allowing investors to build rental apartments *and* allowing them to operate those apartments in a manner consistent with making a reasonable profit.
Remember: Every developer of rentals is either a landlord-in-waiting or hoping to sell to one.

It’s the most wonderful time of the year 🎶
The 2026 AI Engineering Survey is live, this year in partnership with @NotionHQ and @vercel. If you’re an engineer working on AI products, we’d love you to fill it out.
We're raffling off fantastic prizes too.

Finally a semi-useful read on Mythos that is free of myth and talks about what this means more practically (not this is the end of the world as we know it, but how do we deal with faster patches and attacks from AI as other models scale to chained exploits)?
This is the kind of conversation we need, not idiotic ones about the end of all software.
We need "what is the right answer?" because these models are coming and will get better so how to we put our heads together and make better/more secure software across the world?
And it can't just be patching the 100 or so projects that got access to Project Glasswing.
That is not gonna help the world.
We need to figure out how does everyone else who is not part of the special chosen people to get blessed with access to test and patch their stuff, aka the open source projects and closed software that is not Office or Cloudflare but the 99.99% of software that runs everything else in the world?
What is the right loop cycle to help people patch and fix things at the source?
In the long run, AI will make software more secure, not less.
But it will change how teams have to work to get there.
Figuring that out means putting it in more team's hands sooner rather than later.
Cloudflare: Cloudflare's security team spent the last few weeks testing Anthropic's Mythos against fifty of our own repositories. What we learned about offensive AI, why faster patching is the wrong reaction, and what the architecture around vulnerabilities has to look like next.

The bitter lesson in 26 words:
Don’t be distracted by human knowledge, as AI has been historically.
Instead focus on methods for creating knowledge that scale with computation, like search and learning.

My AI Engineer Singapore journey:
Most exciting:
First time traveling abroad after joining http://Z.ai, and my very first English talk on stage.
Most challenging:
Spoke in a three-speaker session alongside OpenAI and Google DeepMind, with http://Z.ai presenting third. Survived, and realized that people did like some of the things I shared.
Most surprising:
Met the @MiniMax_AI team three times, and each time we talked for at least half an hour. Instead of discussing AI, we talked about humans.
Most rewarding:
Learned a lot from deep conversations with @GeoffreyHuntley , @sarahookr, @picocreator, and many other wonderful guest speakers.
Biggest regret:
Did not have enough time to talk with everyone I wanted to know.
Proudest moment:
Stayed until the very end of every event. Got a lot of positive feedback when people had heard of GLM and http://Z.ai. When they had not, it felt just as meaningful to introduce what we are building.
Funniest part:
As one of the major sponsors, somehow did not look like a sponsor at all. No booth, no side events, and not much sponsor presence prepared on our side. More like a group of friends showing up to support the community.
Thanks to @aiDotEngineer, 65 Labs, all the organizers, speakers, builders, friends, and my team who made these days so memorable. I truly want to @ all of you, but I am afraid I might miss someone, so please take this as an @ from my heart.
See you at the next AI Engineer event, and see you again in Singapore soon.

Worth noting that the current datacenter backlash is what 40 years of anti-nuclear activism looks like in the present tense. The same movement that killed clean baseload is now mad that compute is straining the grid. Colossal self-own.
whimsy lohan (in SF): Climate activists have spent decades shutting down nuclear, blocking clean energy transmission, and trying to ban research on cooling the planet... but the movement that wins the next century will be led by the people doing the work, not those protesting it. New essay out today👇

Re @Replit can save you money if you know how to prompt correctly and research your needs. I can now track traffic for all my sites the way I want and need to, without paying someone else. It took me longer to add the tracker info in all the sites than to build this data site.

Doom Just Received One Of The Highest Cultural Honors In The US http://dlvr.it/TSbrMH

Superintendent Su and the Board of Education received more than 2,100 letters from SF parents and the community urging a halt to the unvetted, 2-semester Mandate of "liberated ethnic studies" Voices. Su is being summoned to a congressional hearing on “Breaking Trust: Attacks on Parental Rights, Inappropriate Content, and Legal Abuses in America’s Schools”
https://thevoicesf.org/sfusd-summoned-to-congress-amid-firestorm-over-ethnic-studies-and-attacks-on-parental-rights/
@kunalmodi @DanielLurie

Design with Replit, build taste
James Leong: Replit is my UI/UX designer
As a creative director creating design languages is faster than ever.
Hop into canvas and start building live useable components harnessing Swift, Shadcn/ui to use in any future site ask agent to take it and make it a skill
Taste over everything

will post more about (final) day 2 of @aiDotEngineer Singapore soon but memorable highlights i wanted to remember is how @theesabina and @bytheophana (both from nyc) were genuinely shocked and on guard by how friendly people were at the summit 😅 nyc energy clashing w spicy-brain singapore ai scene energy
other highlights:
- @DelaneSays bringing her 4yo kid to watch the conference!! little ki learning about llms and computer use before algebra, befriending BOTH robots and humans with @waqasmakhdum @DhruvDiddi 😆
- after 15 hours of sleep, first thing i watched was @agrimsingh's closing speech :')) the 65labs team is an inspiration on how to bring SOUL and HEART to a growing scene
- @GeoffreyHuntley giving me great advice on what hats to not be wearing lmao. ty for saving me from getting cancelled by accident && for bringing ur sick hat fashion to the scene
- bringing @convex team @waynesutton @mikeysee @existentializzm on a duck tour in singapore!! it'd be so great to see convex land in singapore more permanently :))
- @yoimnotkesku and @adlinzainal fighting over who gets to pay for the volunteers!! kes rly acing the 'singaporean citizenship application' at every turn haha ty @adlinzainal for being the mom of the team :D
- went on a dinner tour with newly found aie friends and @ktoya_me suddenly wanting to do a mini-hack with @ElevenLabs and us at AIT, then bumping into @latentclaw who was super SUPPORTIVE OF IT lfg
- me realizing there were FOUR AIT ex/current leads in AIE Singapore: @tokengobbler matthiaslubken @Stefania_druga and @Calclavia -- so many community-minded folks doing amazing things! pity i didnt get to meet half of u in person :')
- @unprofeshme @adlinzainal sharing their MAXIMALIST, GIFT GIVING love languages and giving me inspiration for future merch
- bumping into @minu_who and @hrishioa at the afterparty, and the alcohol-fuelled conversations after 😂
- @brianchew & yiming's quiet but consistent presence in providing for volunteers throughout the entire conf. so much wouldn't've been possible without their help
- @rachpradhan planning to post-train a model into making me it's default taste ☠️
- SO MANY inspiring kids who blew me away with their energy and heart, @Shreyansh_Ag1, @hehuabuilds and @PairieK, @wilsonsfh_ @lavanya_g112 -- this scene is meant for you folks and we hope to see u more often here (or building ur own scenes!) :)
- @justinbaird and the tesseract team's lovely heartwarming BCI demo that showed us how tech can meet us where we are, literally :'')
- @swyx taking a photo of us in the backstage before putting it on the last slide of his closing speech :'') mentioning @nathanwangliao as his personal chatgpt lol and just as importantly, his hope on what Singapore could be! that every great thing that happened in singapore started off as a shower thought in someone's mind :)
- @theesabina's speech!! i legit thought she was a theatre kid -- by far the best AIE demo i've gotten to hear and soooo soo many notes to take down on what makes something engaging, fun and technical!
- @mdda123 being one of the OG HEARTs AND SOULs of the ai scene here :) i see you sir and i thank you for the heart work you've done to bring high signal events to folks
- so so many other inspiring folks who have GIVEN so much, and pushed and are continuing to push me to grow as a community lead in ways i never knew were possible
really -- im in awe of everyone i got to meet, no matter on . we are the scene, and we'll keep growing larger from here! watch this space 🫡😎
p.s. @theesabina i did not manage to get a @magicpatterns cap!! they look SO GOOD, pity you had to head off so quickly and we'd love to host u whenever you're back. love to try the tools out soon :)
@adlinzainal @brianchew @gabrielchua @agrimsingh @SherryYanJiang @unprofeshme @NehallAgrawal @aimuggle @65labslah @ivanleomk

There is now irrefutable evidence that China and other foreign actors are heavily involved in the campaign against American AI.
And we’ve got all the receipts.
In this groundbreaking report, we combed through nearly 100 open-source materials to document how international actors are working through state media organizations, nonprofit networks, and dark money groups to shape US policy and public opinion on artificial intelligence.
Bitcoin Policy Institute: NEW RESEARCH from @SamLyman33: China is interfering with US data center development in a campaign to slow the American AI buildout.
In this groundbreaking report, we document 3 vectors of foreign influence: Chinese state media, the Singham network, and foreign billionaires.

Still blown away daily by just how good @Replit is.
The orchestration of models is way beyond what anyone else is doing.

People are unironically protesting AI data centers with AI slop.
Everything is so stupid right now.
Polymarket: NEW: First renderings of Kevin O’Leary’s 60-building, 7,500,000,000-watt Utah data center project have been revealed.

Flock saves lives.
Rahul Sidhu: Last year, the city of Austin turned off their Flock cameras as the result of a targeted misinformation campaign.
This weekend, for nearly 24 hours, three suspects drove around Austin in stolen vehicles, undetected, conducting a shooting spree at 12 separate locations. They

We need more… brave journalists bravely protecting women from the terrifying danger of learning useful skills.
Because nothing says ‘I care about women’ like telling them that picking up AI is class treason and they should stay helpfully helpless instead.
Garry Tan: The NYT is predictably tearing down Reese Witherspoon for encouraging moms to try AI before they ingest the anti-AI pablum as truth
Instead of linking to the NYT op-ed, I think you should watch this video and encourage you to follow Reese Witherspoon on Instagram

>here's how an AI brain actually remembers.
>you toss in notes, toss in decisions, toss in chats.
>AI turns each into a card, drops it in a pool. before answering, it pulls 12 and lays them on a desk.
>once the pool hits 10k cards, AI pulls wrong ones without noticing. holds up a 6-month-old card and tells you, with a straight face, that this is the current rule.
>get this layer right, your card pool gets thicker every day, your AI feels more like someone who actually knows you. and you don't have to wonder if it's reading you an old card anymore.
Vox: http://x.com/i/article/2056447894412996608

CEO and co-founder of Replit, @amasad is building the platform that makes coding accessible to billions, from kids creating school projects to enterprises running their business.
See him take the stage with Spike Jonze on day one of Vibecon.
NYC, June 17–18.
Get your tickets at http://vibecon.ai

The community-installed swing used by @sfgov in a tourism commercial was removed by @sfgov for not being authorized.
Zach Coelius: It’s so sad. The bureaucracy of the City of San Francisco seems to have finally won the war of wills with the guerrilla swing installers of Bernal Heights.
For years an amazing swing would go up and the city would cut it down.
Now it seems down for good. 🥲

A few things to get out of the way:
1. There's no official gbrain coin
2. Don't need funding
3. Don't need help getting followers
✌️

Funding these clowns was such a tremendous fuckup b the climate philanthropy world.

Pelosi isn’t pro-housing. She doesn’t even understand local SF issues on a deep level. She’s been disconnected from SF culture for decades. She relies on a small inner circle here, while almost entirely focused on national DC politics.
Chakrabarti is done as a consequence. Funny
Sheel Mohnot: WTF. Pelosi endorsed Connie Chan, a progressive supervisor in SF who opposed the Chesa Boudin recall, blocked every form of housing, and voted against streamlining permits and cutting red tape.
The opposite of an abundance candidate.

Data centers are not stealing your water