Sabitlenmiş Tweet
ℏεsam
7K posts
ℏεsam
@Hesamation
apes made fire, then GPUs, then ASI
the loss landscape Katılım Eylül 2024
471 Takip Edilen82.8K Takipçiler
This is big guys. a Senior Researcher at Google DeepMind and 600 other employees asked Sundar Pichai not to use their AI models for classified military use.
Google signed the Pentagon deal anyway.
> 2014: Google buys DeepMind on the condition of no military use
> 2026: DeepMind signs a Pentagon contract for "any lawful government purpose"
> 600+ employees signed a letter asking Pichai not to put their AI models on classified networks
> he and potentially other employees found out the deal is signed through press, not even an internal announcement
Andreas Kirsch 🇺🇦@BlackHC
I work at Google DeepMind. This won't make me popular. But it's all public reporting: 2014: DeepMind reportedly sold to Google on conditions: no military use, independent oversight 2026: a Pentagon contract for "any lawful government purpose" Not one safeguard survived intact
English
GPT-5.6 Sol doesn’t allow Fable to leave the paid plans.
Sam Altman@sama
GPT-5.6 sol is half the price and ~twice as token efficient as fable in many cases for accomplishing the same task. happy to deliver at one-quarter of the price.
English
teachers reading this rn:
(Claude is coming for you)
Claude@claudeai
We're introducing Claude for Teachers: free access to premium Claude capabilities for verified K-12 educators in the US, with a library of teaching skills and a direct connection to evidence-based curricula, mapped to academic standards in all 50 states. claude.com/solutions/teac…
English
Demis Hassabis (Google DeepMind) asks the government to step in on frontier AI, or even order a slowdown in AI if risks become high.
"This could apply to Frontier-class models NO MATTER THEIR COUNTRY OF ORIGIN or whether they are OPEN or CLOSED"
with Dario and Demis actively pushing for this framework, I don't think it's really far from seeing a form of regulation for open source AI, or even being allowed to fine-tune a frontier open model and upload it on Hugging Face.
Demis Hassabis@demishassabis
English
ℏεsam retweetledi
Apple’s M7 Ultra chip rumored to be a local AI workhorse as it supports 1.5TB of RAM. for reference, that’s enough to run Kimi or GLM models locally.
AppleTrack@appltrack
Apple's M7 Ultra chip coming in 2029 is rumored to support 1.5TB of RAM. This would make the processor much more capable for on-device AI.
English
> find a boring app people already pay for
> rebuild it as an agent first product
> ship fast
> repeat until one reaches ~$200K/month
GREG ISENBERG@gregisenberg
English
great model,
completely avoidable drama
SpaceXAI@SpaceXAI
We care deeply about your privacy and respect customer choice. For teams using zero data retention, no trace and code data is ever retained. All API key use of Grok Build also respects ZDR. If ZDR is disabled, the /privacy command is available in the CLI to disable data retention, which also deletes previously synced data. Run the /privacy command to view or change your settings at any time.
English
Dan Koe has talked about one person businesses for ages. it’s not a choice, it’s necessary.
but now that everyone can build with AI, your unfair advantage is no longer just building.
this article is a good place to start.
Ronin@DeRonin_
English
just need to mortgage my non existent house and my dream of local ai will crawl forward at two digit tok/s !
yay
ℏεsam@Hesamation
Apple’s M7 Ultra chip rumored to be a local AI workhorse as it supports 1.5TB of RAM. for reference, that’s enough to run Kimi or GLM models locally.
English
> director of design at Figma
> led Claude’s redesign
> led design of Claude Cowork
NOW JOINS CURSOR
@SpaceXAI might be emerging as a real poach player.
jenny wen@jenny_wen
ok! quick updates from me: - i left anthropic (who does that?!) - had a baby (i love her) - and am joining @cursor_ai as head of design (eep!) it's been a low-key dream of mine to nurture a team that cares so deeply about craft, quality, and building great tools. very excited!
English
@FabianFran52130 no YOLO
no running while sleep
no prod
as is the case with any model
English
@Hesamation Well, he's a dumb ass for allowing it on aproduction system, typical amatuer hour lol.
English
@satellitedown the fact that neither provided their prompt and setup (even after being asked so much) is sus, but nonetheless running gpt (or any AI model but this is reported much more recently with Sol) in YOLO mode and unsupervised is a recipe for disaster.
English
@Hesamation both of these cases were user error and then they dramatized it for likes. sol is no more likely to delete stuff than any other model.
English
Fable/Mythos has been a carrot on a stick since day one:
> Mythos for only trusted partners
> Fable released as substitute
> heavy guardrails for anything
> government takes down Fable
> Mythos still out of reach
> Fable deadline extended again
> then extended another week
DAMN
Claude@claudeai
We're extending Claude Fable 5 access on all paid plans, as well as keeping Claude Code’s weekly rate limits 50% higher, through July 19.
English
