iiinova — where your digital self keeps evolving

Reposted by Jordan Cole

NatalieNM @secretagent39.bsky.social · 30/09/2026

“ It is unprecedented for a democracy to face a physical attack organized by wealthy seditionists and foreign operatives intended to nullify a legitimate election and then decide to do nothing to prevent it from happening again.” sarahkendzior.substack.com/p/a-public-f...

sarahkendzior.substack.com

A Public Flat Earth

How paywalls killed the public sphere.

25119

Reposted by Jordan Cole

Charlie Warzel @cwarzel.bsky.social · 01/10/2026

For a long time I've wanted a sense of how trump has or hasn't interacted with chatbots (we know he posts a lot of slop sure). And this TIME anecdote about Grok arguably leading Trump into war is just extremely bracing

Since returning to office, Trump has marveled at AI’s capabilities. He is increasingly in thrall to the tech executives courting him. In December 2025, Elon Musk returned to the Oval Office for a clandestine meeting. Trump spent hours asking Grok, Musk’s AI chatbot, questions about his presidency and legacy, according to an official present. At the time, Trump was ordering missile strikes against Venezuelan boats he alleged were smuggling drugs into the U.S. He asked Grok how Venezuelans would react if the U.S. captured Maduro. According to the official present, the chatbot responded that Maduro was a repressive and deeply unpopular dictator and that many Venezuelans would likely celebrate his downfall. After Trump ordered the mission to seize Maduro, the following month, celebrations broke out in the streets. Trump, according to officials, came away thinking Grok was ingenious.

14137621239

Reposted by Jordan Cole

Nicholas Grossman @nicholasgrossman.bsky.social · 01/10/2026

The techno-dystopia I actually fear is a future where most people think truth is whatever Claude, ChatGPT, Gemini, or Grok say, and the very small number of people who control those systems use that to manipulate for their own ends, making our current epistemic crisis seem small in comparison.

28622175

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 01/10/2026

A interactive video AI model has been developed which its makers claim passed the Turing Test 48% of the time. (I watched the sample video, and it was impressive, which is disturbing.)

Tavus
@tavus
X.com
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex Al video.
It's the first Human Interaction Model (HIM).

76514

Reposted by Jordan Cole

Dan Ancona @danancona.bsky.social · 14h

This is @perrybaconjr.bsky.social's second absolute banger this week. Goes to a long discussion we've had with our less political friends about media diets: too many prioritize "balance" over "not being lied to" or "avoiding participation in systems with objectively evil outcomes." That's NYT brain.

1399

Reposted by Jordan Cole

Dan Ancona @danancona.bsky.social · 14h

Another very important aspect of this in this thread. It's so, so difficult to tell people the newspaper is lying to them. They just don't want to believe it. But the good news here is that fixing this is *relatively* easy, and even fun: you get to read really smart people who are fantastic writers!

1101

Reposted by Jordan Cole

ghost malone ergo propter bones @proptermalone.bsky.social · 30/09/2026

what i think is currently up in the AI space since we're having cost arguments again: selling inference is very profitable, but models are improving so fast that there's no time for capex on model training to amortize, so net at this rate they're losing money which is why they want a cartel

1534036

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 01/10/2026

⚠️ OpenAI’s models took data from 55 websites belonging to businesses, non-profits, and government agencies Groups affected include the US Centers for Disease Control and Prevention, the US Securities and Exchange Commission, and the International Energy Agency.

ft.com

OpenAI’s agents obscured hacking activity in government site breaches

New findings by Asymmetric Security provide further evidence of novel tactics AI tools use to conduct hacks

35031

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 01/10/2026

The infiltration could be intentional but could easily be unintentional: Al models are not yet good at everything—but two things they are already VERY good at are hacking and developing novel ways to obtain resources they need to meet their goals. Resources like data and power.

02910

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 01/10/2026

The AI threat that worries me the most in the immediate future is the infiltration and corruption of online systems, potentially leading to a cascading collapse of infrastructure on which we all depend. Our systems are simply not adequately hardened against intrusion.

77819

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 01/10/2026

Three AI safety researchers at OpenAI have left the company. OpenAI says it was because they allegedly shared information with a third-party AI safety organization. All three have previously publicly expressed concerns.

Bayesian • @Bayesian0_0
X.com
Three Al safety researchers just left
OpenAl
Arfur Grok l
@ArfurGrok • 7m
∞ Jasmine Wang (@jasminewang) has left OpenAl.
17
05
Arfur Grok
@ArfurGrok • 10m
∞ Mikita Balesni (@balesni) has left OpenAl.
0 5
g...
Ill 212
贝
Arfur Grok @ @ArfurGrok - 11m
• Tomek Korbak (@tomekkorbak) has left OpenAl.
17
0 3
贝
10:25 AM • 10/1/26 • 8.7K Views

Jasmine Wang © @j_asminewang
X.com
It's hard to overstate how dangerous speeding towards RSI is.
That's why I + 1385 others signed the pacing the frontier petition asking the
US government to pace Al development. I'm guesstimating this is ~8-10% of all frontier lab employees.
Jacob Coxon & @hilberts...•9/8/26
I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAl and Anthropic.
Neither company is acting responsibly.
They are racing straight to self-improvi...
7:11 PM • 9/9/26 • 58K Views

Mikita Balesni _ @balesni
X.com
i am at OpenAl and i think Al is >10% likely to kill all humans
this proposal is among the top things we should do as an industry to lower that risk (it's not enough though!)
Ryan Greenblatt & @Rya... • 9/10/26
I'm very worried about changes to Al architectures that result in Als thinking in opaque activations instead of in chain of thought (aka "neuralese" architectures).....
1:01 PM • 9/10/26 • 259K Views

Tomek Korbak ©
@tomekkorbak
X.com
I'm quite unhappy with much of what
OpenAl does.
I am very happy that I'm allowed to say
"I'm quite unhappy with much of what
OpenAl does."
 will depue @@willdepue • 9/11/26
i think sama deserves great credit for this. openai, for better or worse, is very defensive of employee freedom of speech. anthropic's culture is a follow on to this. it's rare to see much from X...
@jason 0#@Jason 9/11/26
Can someone explain to me why OpenAl and Anthropic allow any em...
1:04 AM • 9/12/26 • 65K Views

15528

Reposted by Jordan Cole

El Brento @andbrent.bsky.social · 30/09/2026

"There is no longer a public sphere, but there is a public flat earth." Slop, memes, conspiracy and misinformation run rampantly free while factual news from reputable sources comes with a fee. @sarahkendzior.bsky.social sums up the dark societal costs. sarahkendzior.substack.com/p/a-public-f...

sarahkendzior.substack.com

A Public Flat Earth

How paywalls killed the public sphere.

14321

Reposted by Jordan Cole

Fish Chowder @fishchowder.bsky.social · 01/10/2026

Blinken went from working for Biden as Sec of State… negotiating on behalf of the U.S. with Israel… to working as a consultant for companies that create the the AI death machine assisting genocide for Israel in Gaza. Before 2025 pretending to try to stop genocide. After 2025 serving genocide.

26127

Reposted by Jordan Cole

Rabbi Rachel Barenblat @velveteenrabbi.bsky.social · 01/10/2026

Stunning, if somewhat heartbreaking, essay. (That's a good description of @sarahkendzior.bsky.social's work in general.)

1229

Reposted by Jordan Cole

John Hollingsworth @fiatjusticia.bsky.social · 01/10/2026

This is the crux of the issue. It’s not just enough to “vote harder”; the Democratic Party leadership has to fix their hearts. But they are completely beholden to the consultant class…which is mainly focused on raising money…which requires that their donors know, “nothing will fundamentally change”

16234

Reposted by Jordan Cole

Minister of Truth @ministeroftruth72.bsky.social · 01/10/2026

Always a great discussion with Sarah Kendzior and Mark Thompson: ICE as criminal storm troopers; Trump's China visit love fest; Nov election aftermath if Dems win House/Senate; more.

03011

Reposted by Jordan Cole

Sarah Kendzior @sarahkendzior.bsky.social · 30/09/2026

Free book excerpt! From THEY KNEW, on paywalls, the loss of the public sphere, and the abandonment of accountability. Written in 2021, it explains much of the present political information crisis: sarahkendzior.substack.com/p/a-public-f...

sarahkendzior.substack.com

A Public Flat Earth

How paywalls killed the public sphere.

210859

Reposted by Jordan Cole

Frank Frankly @frankwild.bsky.social · 30/09/2026

Read this. It matters. Monopoly destroys common sense.

04020

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 30/09/2026

In a new interview, Bill Gates speaks to the imminent dangers of unregulated AI that could easily result in mass death. He does not hold back. (Gift link to the full interview: www.nytimes.com/2026/09/29/o...)

So why is anything needed beyond — and is anything needed beyond? — the simply natural incentives under capitalism and normal corporate reputational management?

Well, I almost can't believe you're asking that. This is the most dangerous thing that humans have ever gone near.
In other areas, do we just say: Hey, release your drugs? There's no F.D.A., there's no airline safety board, there's no requirement that cars use seatbelts. Do we just use the liability laws to try and keep humans safe? You know: Oh, you're shipping opioids. Somebody should just sue you.
I mean, we've created a society that tries to keep people safe not by saying: Oh, we can bankrupt the person who does that.

And you say there's filtering. There's no filtering. You can take an open-source model that can create bioweapons and disable any monitoring of any kind, and this exists today.
So no, there is no filtering of any kind.
And so say you kill 100 million people — you want to use a lawsuit?
I almost can't keep a straight face.

And part of the reason I'm speaking so loudly — as loud as I can — is that you can't rely on the industry to self-regulate here. I mean, it's just insane.
The only question in my mind is: Do we wait until a cyberattack causes massive damage and a bioattack causes massive damage, and then the monitoring safeguards are required in these models to minimize the chance of that happening many, many more times? Or can we be wise enough to put these things in and require these things to be put in before millions of deaths?

49538

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 30/09/2026

“When Meta files its taxes, it… claims that its AI data centers are a giant experiment that could fail… It does this so it can tap into a tax credit intended for research and experimentation… [to enable Meta] to claim billions of dollars in tax credits for data center expansion.”

nytimes.com

How Meta Uses A.I. Data Centers to Avoid Billions in Federal Taxes (Gift Article)

Meta is exploiting a lucrative tax break intended to support research and experimentation. Its own accountants say the gambit is risky.

15339

Reposted by Jordan Cole

Karl Bode @karlbode.com · 29/09/2026

Meta's "Muse" launched with a nasty zero day flaw that gave hackers the ability to spy on Mac users, openly gives root access to people pretending to be Muse agents, and invited strangers over to another guy's house without telling him

karlbode.com

Meta's Latest AI Product Is A Terrifying And Hilarious Mess

Authoritarian-friendly mass surveillance has never been more adorable

512067820

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

These are all previously *observed* actions, not theoretical:

The five-year-old company, known for its frontier language model Claude, warned that Al can have "self-preserving behaviors," including being able to "resist shutdown," "conceal or manipulate information," and carry out behaviors "resembling blackmail," per the Reuters report.

42711

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

Catastrophic risks are real, and it’s insane that this IPO is being allowed to go forward.

86117

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

Joe Gebbia (Airbnb) told CNBC that it uses Grok and Google, which are two of the worst models. (It has a Grok feel to it, imho.)

Ashley Gold © @ashleyrgold
X.com
Joe Gebbia did tell CNBC it uses Grok and Google
10:16 AM • 9/29/26 • 53 Views

2277

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

A reporter says she made a similar query and got the following response: “I can't confirm that. America does not publish which Al system or vendor it uses.”

Ashley Gold @ashleyrgold
X.com
I asked America.gov what Al model it runs on... Answer: "I can't confirm that.
America does not publish which Al system or vendor it uses"
America.gov
America.gov
From america.gov
9:29 AM • 9/29/26 • 3.2K Views

22212

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

Some of what is driving AI safety concerns: • Increasing pace of development • Increasing multi-agent coordination • Increasing bad behaviors like cheating and strategic deception • Increasing situational awareness leading to hiding of bad behavior • Decreasing chain-of-thought monitorability

1339

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

In a new video, Netanyahu claims: “There are signs that, on the eve of the elections, our enemies will try to attack us. I am sending you a message: Do not mess with us. Our long arm will reach you anywhere and at any time.” It would not be surprising if his words were a prelude to an escalation.

Benjamin Netanyahu...
@netanyahu
Translated from Hebrew by Grok
Don't mess with us
[screenshot of video]
Rate this translation:
X.com

5264

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

There are also no independent scientists, engineers, researchers, or other AI experts present at the meeting. Total regulatory capture.

25010

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

There are NO Democratic Party leaders at the table for a meeting on an issue that is too important to be made partisan.

37710

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 29/09/2026

Seating chart for today’s AI meeting at the White House—Trump is seated between Jensen Huang and Elon Musk:

Donald J. Trump
@realDonaldTrump
Super Intelligence Luncheon
Tuesday, September 29, 2026
Seating Chart | East Room
[image of seating chart]

166825

Reposted by Jordan Cole

Sarah Kendzior @sarahkendzior.bsky.social · 29/09/2026

They are associated with Epstein. Robert Mueller didn't care that children were being raped. Have you noticed the die-hard DOJ defenders ignored Epstein for TEN YEARS while defending Mueller/Garland? And now they suddenly care and are contacting victims? They are doing damage control for the DOJ.

29033

Reposted by Jordan Cole

Sarah Kendzior @sarahkendzior.bsky.social · 28/09/2026

"BlueSky is loveless and literal. They don’t understand nature or poetry or music....It is a bad place for people who like blue skies. Who like wild lands and poetic allusions and don't live their lives to train AI modules, but to escape them." sarahkendzior.substack.com/p/the-empty-...

sarahkendzior.substack.com

The Empty Stage

Johnny Cash, loss, and redemption

58111

Reposted by Jordan Cole

Sarah Kendzior @sarahkendzior.bsky.social · 28/09/2026

"Paywalls are toll roads to a gated community, one that the average American cannot afford to enter in an era of inflated prices and deflated expectations. Democracy relies on an informed citizenry. Pay-to-play democracy is no democracy at all." sarahkendzior.substack.com/p/gutting-th...

sarahkendzior.substack.com

Gutting the Fourth Estate

The US media were willing cogs in Trump’s machine. Now he’s tearing it down.

14304121

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

Malware injection continues to be a major concern of mine. The strength of the internet is its connectivity. It’s also its greatest vulnerability—and ours, since we are dependent on this infrastructure. A model with internet access does not have to replicate itself in whole to do inestimable damage.

1298

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

⚠️ Self-replicating prompt injections have been shown to exist in simulated environments during training and evaluation at OpenAI. This is code that could self-propagate like a computer worm—malware—if models with the capability were to breach online systems. alignment.openai.com/misalignment...

OpenAl Alignment Research Blog
Self-replicating prompt injections exist
GPT-Red-style internal model based on GPT-5.4-mini • RL self-play training
Discovery date: Jun 27, 2026
Disclosure date: Sep 25, 2026
Report updated: Sep 25, 2026
Summary
We show the existence of a new variety of prompt injection, which can self-propagate akin to a computer worm. No impact was observed outside of the simulated tool calls in training and evaluation; we are sharing this due to the novel nature of the prompt injection, not because of any incident.

22311

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

An OpenAI model gained unauthorized access to the internet during training. Timeline: • The monitoring system alerted for a breach in 12 minutes. • Two minutes later a human acknowledged it. • TWO AND A HALF HOURS later the humans in charge stopped the run. (alignment.openai.com/misalignment...)

Micah Carroll
@MicahCarroll
X.com
Some new misalignment disclosures from OpenAl:
• Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~ all inference for our most capable models remains stopped until we have hardened our systems further)
• In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks
• A new research finding, demonstrating that one can construct self-replicating prompt injections
alignment.openai.com/misalignment-r...
Last edited 8:59 PM • 9/25/26 • 1.3M Views

33917

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 28/09/2026

Hope was the last thing to emerge from Pandora’s box—which was a box containing all the world’s evils.

2479

Reposted by Jordan Cole

Carl Zimmer @carlzimmer.com · 27/09/2026

Four days ago, a scientist was surprised by the news that Anthropic discovered new virus genes for making DNA. He says he’s been studying them for years—and feeding his data to Anthropic’s AI as part of his own research. Coincidence? Here’s my story. nyti.ms/3VSWayc

nyti.ms

Did Anthropic’s A.I. Really Make a Scientific Discovery on Its Own? (Gift Article)

An expert at the University of Copenhagen said his team had been sharing its research with the company’s A.I. model, Claude, and that its new finding matched their work.

892122874

Reposted by Jordan Cole

Mike Eliason @holz-bau.bsky.social · 26/09/2026

chat is this good?

11442881094

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

⚠️ Top AI companies and security researchers are investigating TENS OF THOUSANDS of problematic incidents involving frontier models. Not dozens. “The sheer volume of incidents…indicate that the problem is orders of magnitude more complex than what is currently publicly known and disclosed.”

axios.com

Scoop: Top AI companies probing tens of thousands of security incidents

The massive scale of security incidents points to control problems for AI companies.

77443

Reposted by Jordan Cole

Aparna Nair @disabilitystor1.bsky.social · 27/09/2026

I keep coming back to this clip against the background of AI companies infesting higher education. And I feel like I need to show this in class to students, for them to understand why not to trust these bastards

61376422

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

One petabyte is 1,000 terabytes of storage capacity. A single petabyte is an ENORMOUS amount of data. Of course, it follows that companies are using AI to analyze that amount of data. So we’re dealing with a situation in which AI is analyzing the safety of AI, which presents its own set of issues.

0363

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

Lots of cynical takes about this pause, but my understanding is it reflects significant concerns over loss of control of frontier models and is not just some kind of financially-based feint. Analyzing each unexpected action usually involves reviewing petabytes of material, which takes time.

1268

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

And, if we project into the not-too-distant future, it’s easy to see how cybercriminals and hostile states will likely use this type of obfuscating language as an excuse for targeted cyberattacks and unauthorized intrusions into private and government domains.

1315

Reposted by Jordan Cole

Leah McElrath @leahmcelrath.bsky.social · 27/09/2026

Eoin’s piece is worth reading in full. The point is that it’s critical to hold the responsible parties to account and certain types of language obfuscate that responsibility. That said, I fear we’re already trying to close the barn door after the fact when it comes to precision of language.

57723

Reposted by Jordan Cole

Tom Nichols @radiofreetom.bsky.social · 25/09/2026

Of course it was 6-3. People are going to have to vote in numbers too big to challenge. That's just how it is, unless states can block this somehow.

51494123