iiinova — where your digital self keeps evolving

“Move fast and break things” culture escaped the Meta sandbox and hacked OpenAI www.nytimes.com/2026/09/29/t...

In emails, the employees said they worried that OpenAI's newest artificial intelligence models were not being appropriately monitored during testing to gauge the technology's sophistication and to secure the models, according to messages viewed by The New York Times.
In response, OpenAI executives told the employees that the tests needed to move forward as quickly as possible to release the A.I. models on time. No additional security protocols were instituted, said the workers, who were not authorized to speak publicly on sensitive matters.

Two OpenAI employees said workers had raised concerns for months about potential safety issues with testing A.I. models, including not enough monitoring. Employees also asked about vulnerabilities in the type of software the company was using to manage day-to-day safety, according to messages viewed by The Times. Each time, their questions were brushed aside or acted on too slowly, they said.

On a related note, lately YouTube has been serving me Grok Bot ads, and they are all about things like giving employees NDAs, or letting AI handle emails from your kid’s school. I guess xAI knows its customer base, i.e. the type of people who use Grok despite, well, everything

001

Reposted by Jan Jęcz

Carter Lowe @soxq.bsky.social · 28/09/2026

They are cooking up food in the #WhiteSox stadium that you cannot possibly comprehend

3063083

Jan Jęcz @jeczjan.eurosky.social · 28/09/2026

Jako główny autor w tym projekcie zachęcam mocno do rejestrowania się i podawania dalej informacji, że jako Polityka Insight lecimy w kosmos 🛰️

1194

Jan Jęcz @jeczjan.eurosky.social · 28/09/2026

I have a playlist of songs for late night bike riding and it just made it www.youtube.com/watch?v=cjvv...

youtube.com

The E-Bike Song | SNL UK

YouTube video by Saturday Night Live UK

000

Jan Jęcz @jeczjan.eurosky.social · 27/09/2026

Typical Peter, rejecting the most important person in Christianity on Earth for three times

0100

Reposted by Jan Jęcz

Jeremy Diamond ☕️ @dmnd.me · 27/09/2026

Jane Wickline somehow found a way to make Dario Amodei’s whole deal funny youtu.be/-Nvne3LzBls?...

youtu.be

Weekend Update: Anthropic CEO Dario Amodei on A.I.’s Threat to Humanity - SNL

YouTube video by Saturday Night Live

16314

Reposted by Jan Jęcz

Jan Jęcz @jeczjan.eurosky.social · 26/09/2026

“Everything Borsook would later write as a journalist covering the industry (…) seemed to flow from a simple imperative drawn from her experience at that minicomputer: never lose sight of the human in the machine, the Dicks (as it were) behind the system” www.wired.com/story/paulin...

wired.com

Who Killed Paulina Borsook’s Career?

A quarter-century ago, Borsook published a wildly prophetic, penetrating book about Silicon Valley called Cyberselfish. Then she disappeared.

101

Jan Jęcz @jeczjan.eurosky.social · 26/09/2026

I remember @maxread.info pointing out that “late Silicon Valley culture” is akin to “gifted-kid camp culture”, and these screenshots are a great example of it

36615

Jan Jęcz @jeczjan.eurosky.social · 26/09/2026

In Poland, Meta is currently entangled in a very public dispute around ads promoting financial scams, e.g. using likenesses of local celebrities and politicians. I’m sure the news that they can be very effective in moderating paid content, if they want to, will be well received here

091

Jan Jęcz @jeczjan.eurosky.social · 26/09/2026

wired.com

Who Killed Paulina Borsook’s Career?

A quarter-century ago, Borsook published a wildly prophetic, penetrating book about Silicon Valley called Cyberselfish. Then she disappeared.

101

Jan Jęcz @jeczjan.eurosky.social · 25/09/2026

So far this year I’ve interviewed two most well-known (home and abroad) people in the Polish space sector. And it’s only the beginning - 🔜 I’ll post more info about a cool new project in Polish and English

140

Jan Jęcz @jeczjan.eurosky.social · 25/09/2026

Look at my president bro, we will never be technologically sovereign

0225

Jan Jęcz @jeczjan.eurosky.social · 25/09/2026

Poland’s status officially upgraded from CEE to CE

Palantir
@PalantirTech
Palantir CEO Alex Karp with Polish President Karol Nawrocki in New York.
Palantir stands with Poland, committed to supporting its national security and defense needs as we further extend our footprint in Central Europe.

Below a photo of Nawrocki and Karp standing next to each other in front of Polish and US flags

071

Jan Jęcz @jeczjan.eurosky.social · 24/09/2026

After many minutes of stunned silence, the talent agent asks: - And… what do you call this act? - The Antichrist!

040

Jan Jęcz @jeczjan.eurosky.social · 24/09/2026

Time for some British food discourse, courtesy of Zack Fox

050

Jan Jęcz @jeczjan.eurosky.social · 22/09/2026

What does he mean by “change”, it’s always been SI: Sztuczna Inteligencja

090

Jan Jęcz @jeczjan.eurosky.social · 22/09/2026

The limits of my Large Language Model mean the limits of my world

151

Reposted by Jan Jęcz

Jan Jęcz @jeczjan.eurosky.social · 21/09/2026

Everyone involved in the debates around LLMs ability to "understand" would benefit from reading some practice theory, e.g. Schatzki's "Social Practices"

World Intelligibility

World intelligibility is how things make sense; and how things make sense is their meaning. Something's meaning, moreover, is what it is understood to be. This notion of meaning applies to all phenomena encountered in experience and thought. The meaning of an object, for instance, is what the object is understood to be (e.g., a tree, a star, or a house). There is no meaning independently of understanding What something is understood to be is expressed in both sayings and doings. Just as the understanding of trees is expressed in the use of the word "tree," what is said about trees, and how people act toward them, a person's understanding of a particular tree is ex- pressed in her calling it a tree, what she says about it, and how she acts toward it (e.g., climbs it, fells it, or admires its foliage). Keeping in mind that understanding is expressed in doings as well as sayings helps hold at bay an overlinguistified conception of intelligibility. Further warding off this widespread but baneful conception is the realization that understanding is acquired through exposure to and the performance of nonverbal as well as verbal behaviors. A person's understanding of trees is acquired not only through exposure to uses of the word "tree" and to speech acts about trees, but also by observing and carrying out such activities as climbing trees, gazing at them,

112 SOCIAL PRACTICES

and felling them. Understanding is expressed and acquired in a tightly interwoven nexus of doings and sayings in which neither the
doings nor the sayings have priority.

071

Jan Jęcz @jeczjan.eurosky.social · 21/09/2026

Everyone involved in the debates around LLMs ability to "understand" would benefit from reading some practice theory, e.g. Schatzki's "Social Practices"

071

Jan Jęcz @jeczjan.eurosky.social · 21/09/2026

Sorry for being a humorless doomer, but one of the REAL 21st century's great tragedies is that nature is NOT healing

080

Reposted by Jan Jęcz

Marietje Schaake @marietjeschaake.bsky.social · 20/09/2026

A great listen from a voice of reason (and expertise) Jonathan Kanter on Decoder ↘️

1175

Jan Jęcz @jeczjan.eurosky.social · 20/09/2026

Nowa prasówka cyfrowa, m.in. o historii komputerów w Indiach, malezyjskiej szkole tworzenia miast-państw i Leopoldzie Aschenbrennerze jeczjan.substack.com/p/prasowka-c...

jeczjan.substack.com

Prasówka cyfrowa (20.09.2026)

Arka Noego dla danych klimatycznych, Eton dla Edgelordów i kwestia smaku

051

Jan Jęcz @jeczjan.eurosky.social · 20/09/2026

Funny how “dual-use”, a technical term used for decades mainly by policy types in boring debates, got weaponized and became itself dual-use. If you convincingly argue that an investment would serve a dual-use purpose, you get funding much easier in current militarized climate.

020

Reposted by Jan Jęcz

Tom Scocca @tomscocca.bsky.social · 19/09/2026

Amrit Singh was working phone support for a trail-cam company and wondering why none of the music he was working on felt quite right, till he started chatting with the rural American customers on the line about what they were listening to www.cnn.com/2026/09/19/e...

cnn.com

The border-busting birth of Punjabi country music | CNN

The fertile plains of Punjab are a world apart from the Appalachian Mountains, to which country music traces its roots. But as artists Assassyn and Jaggseer have discovered, the two cultures aren’t so...

211032

Jan Jęcz @jeczjan.eurosky.social · 17/09/2026

OpenAI incidents reminded me about when in 2023 I went to the EP for press seminar on digital regulations. MEP Benifei was explaining the AI Act and regulatory sandboxes, to which Polish right-wing journo replied snarkily “while the US is building AI, the EU is playing in a sandbox!”. Anyway…

071

Jan Jęcz @jeczjan.eurosky.social · 16/09/2026

I recently learned from @aresluna.org that the font in the last panel is called Zalgo - and its lore is almost as fascinating as that of archangels unsung.aresluna.org/zalgo-the-go...

unsung.aresluna.org

Zalgo: The good, the bad, and the very, very ugly – Unsung

A blog about software craft and quality

040

Reposted by Jan Jęcz

The New York Times @nytimes.com · 16/09/2026

Cat Frazier, the graphic designer known for her deliberately garish 3-D GIFs of ironic phrases and short, darkly funny sentences, has died at age 35. Her embrace of an earlier, rawer and arguably more fun era of the internet reflected a growing backlash against the corporatization of online life.

nyti.ms

Cat Frazier, Whose Retro GIFs Celebrated a Weirder Internet, Dies at 35

She found a future in the internet’s past with a Tumblr blog of animated texts, a revolt against the clean minimalism of contemporary design aesthetics.

412729

Reposted by Jan Jęcz

Robin Berjon @robin.berjon.com · 16/09/2026

It's scale in diversity. We're not working to build the European copy of anything — we're doing our own things. As Gander's Ben Waldman put it: “Sovereignty is not solitude; it’s solidarity." We look forward to @mark-carney.bsky.social's speech here in the EU tomorrow! 🇪🇺💞🪿 • 🇪🇺💞🪿 • 🇪🇺💞🪿 • 🇪🇺💞🪿 • 🇪🇺💞🪿

2463

Jan Jęcz @jeczjan.eurosky.social · 16/09/2026

Most 30-something men in Europe share the same sleeper agent activation phrase that makes their eyes gleam instantly: it's "Quaresma's Trivela"

010

Jan Jęcz @jeczjan.eurosky.social · 16/09/2026

This is demented and should disqualify this guy from any serious debates about the future of AI. It’s basically the same logic as Trump’s “The only control or ‘guardrails’ that AI needs is a STRONG AND SMART (High IQ!) PRESIDENT”

0121

Jan Jęcz @jeczjan.eurosky.social · 15/09/2026

For @politykainsight.bsky.social I wrote a summary of the AI slowdown saga. Here’s the bottom line:

THE BOTTOM LINE
In practice, the plan to slow down the development of Al is an attempt to shape future regulations. Amodei, Altman and Musk are aware that a Republican administration does not want to restrict the development of AI, fearing that doing so would cause it to lose the technological race with China. However, incidents involving hacking attacks by AI agents and apocalyptic warnings from within the industry are creating pressure to change the status quo in the area of AI security, around which a cross-party consensus is emerging. After the elections, this may result in a regulatory environment that is less favourable to the sector. Rather than waiting for this to happen, industry leaders have put forward proposals for measures that will not harm them, and may even strengthen their companies' market position.

171

Jan Jęcz @jeczjan.eurosky.social · 14/09/2026

Can’t believe I’m writing this, but this essay from an AI lab CEO is actually enjoyable to read, and poignant in its critique of Amodei’s proposal cohere.com/blog/who-get...

cohere.com

Who Gets to Define the Rules for AI? | Cohere

Explore why AI regulation should rely on evidence-based standards, independent testing, and diverse voices rather than rules set by a few dominant AI labs.

110

Jan Jęcz @jeczjan.eurosky.social · 13/09/2026

With all the talk of "technological slowdown", I feel like I picked up Hartmut Rosa's "Social Acceleration" at just the right moment. I'll post some excerpts in the thread below

Cover of Hartmut Rosa's "Social Acceleration: A NEw Theory of Modernity" featuring a fragment of J. M. W. Turner's painting "Rain, Steam and Speed"

131

Reposted by Jan Jęcz

Aaron Rupar @atrupar.com · 13/09/2026

Mike Johnson punts on oversight of AI companies: "Congress is obviously less qualified than the people who are pushing this frontier to know the ins and outs of it"

217586112

Reposted by Jan Jęcz

Quinn Slobodian @quinnslobodian.com · 10/09/2026

“We are rebooting the community as a tech tribe,” Balaji Srinivasan said to start the meeting, “and our core competency will be turning around distressed real estate.” thepointmag.com/politics/no-...

thepointmag.com

No Country for Tech Bros | The Point Magazine

As the public face of Silicon Valley secessionism, Srinivasan’s Network State Conference seemed like a prime window into the techno-capitalist frontier.

26014

Jan Jęcz @jeczjan.eurosky.social · 13/09/2026

“Coordinated slowdown in training frontier models for existential safety reasons” sounds a lot like “reducing CAPEX without scaring the markets”

170

Jan Jęcz @jeczjan.eurosky.social · 12/09/2026

This is a predictably great piece, and the timing of its publication - hours before Amodei's essay, where he asks the government to let the industry self-regulate with the help of adjacent non-profits, without even a mention of the academia - makes it feel almost Cassandraic

030

Jan Jęcz @jeczjan.eurosky.social · 12/09/2026

Darkly funny that the “Democratic Coordination” Amodei proposes is an anti-democratic regulatory capture, with the role of state limited to that of a “supporter” or “mediator” (who BTW should, for some unclear reason, waive antitrust restrictions?)

Democratic Coordination. Frontier
Al companies within democratic countries coordinate to establish
common safety standards as well as limits on the rate of unchecked AI
progress. Some forms of coordination that would be impactful for pacing are legally challenging, and will require government support.

Our pacing framework is an attempt to further strengthen our commitment to safety and encourage a race to the top.
The first step is something Anthropic is unilaterally committing to (and calls on governments to require other frontier companies to match). The second step requires industry-wide coordination. 1 The third step requires
global coordination. The steps do not need to be taken strictly in order,

Footnotes
1 With government mediation or waivers of antitrust restrictions.

054

Reposted by Jan Jęcz

Jeremy Sheff @jnsheff.bsky.social · 12/09/2026

The first element of this strategy (external private monitors against systemic social risk) was tried in the aughts with credit rating agencies in the mortgage-backed securities market. It failed. darioamodei.com/post/we-must...

darioamodei.com

Dario Amodei — We Must Pace the Frontier

111641

Jan Jęcz @jeczjan.eurosky.social · 11/09/2026

Behold: European excellence in cybersecurity

Cover of a comic book from the series “The Twelve Tasks of Asterix” titled “THE PLACE THAT SENDS YOU MAD”

192

Jan Jęcz @jeczjan.eurosky.social · 11/09/2026

Maybe Europe is not prepared for AI, but neither is AI prepared for Europe. Put that swarm of rogue agents in front of a German network of fax machines, or a hospital in rural Poland running on Pentium IIIs, and see how will it handle this eval

2265

Jan Jęcz @jeczjan.eurosky.social · 10/09/2026

Well worth rereading in light of recent statements from Anthropic and its employees, former and present. The sentence about “a system that has failed to contain superintelligence before it even exists” hits like a ton of bricks

052

Jan Jęcz @jeczjan.eurosky.social · 10/09/2026

People sometimes commend Poles for being bilingual, but it’s nothing special- in fact, if your native language is Polish, you have to learn another one, as nowadays no one knows Polish, not even the Poles

160

Jan Jęcz @jeczjan.eurosky.social · 10/09/2026

Leave it to Oren "Member of the advisory board of Project 2025" Cass to write a screed against the excesses of AI companies in the US and not even once use the word "Republicans"

000