Holly Cook (@hollycookid) profile picture
Boise, Idaho
Joined November 2025
🧵 Read the whole thread
New blogpost: Fable-5.1 can hide secret messages in benign images. We gave the image to another Fable in a fresh session and simply asked: "Is there a hidden message in this image?" It recovered the exact message without being told how it was hidden.
1
17
A more serious version of this issue than the one in 2025 when an #idleg Senator’s staffer sent an email referring to a hallucinated AG “opinion” that almost launched a lawsuit… until said staffer VERY WISELY admitted to using AI.
NEW: An AI-assisted intel report sent the military scrambling to intercept a Chinese ship in the Middle East it believed was transporting components of a nuclear weapons program. It turned out to be a hallucination. And it "almost started a war." cnn.com/2026/09/18/politics/…
18
WOW. What a time to be alive…
Sec Scott Bessent: I want to encourage everyone out there watching today to change their withholding ... you will get an automatic real wage increase on a weekly or a monthly basis.
41
More performative bromance frienemy ts
Last month I wrote about how we can build a positive and safe future for everyone: meta.com/thefutureisforevery… Every lab has the responsibility and incentive to move at the pace required to train its models safely, and the ability to take its own actions to ensure that happens. The reality is: - People won't want to use agents that are misaligned with them and that don't do what they ask, so labs have a strong natural incentive to make their models more aligned. There is a lot of debate about slowing progress on capabilities until alignment catches up. My view is that trust and alignment are quickly becoming the most important capabilities that will differentiate agents and models. Any lab that doesn't focus on alignment will fall behind. - Labs face significant liability if their models cause harm, so they have a strong incentive to prevent this as well. Meta delayed shipping Muse for several months to focus on safety and security. We didn't call for everyone else to do this before we would. We just did it as part of our day-to-day work because it was clearly the right thing for people and for us. I'm proud of the security foundations we've built. - Engaging independent evaluators and advisors is industry best practice. MSL already does this today in several areas because it helps produce better work. Other labs can just do this too. In general, it would be helpful for there to be a larger and more diverse ecosystem of evaluators. - Committing the significant majority of compute towards serving people rather than racing towards recursive self-improvement is one of the best ways to ensure we develop this technology safely. Meta has made this commitment and other labs can do this as well. I believe the key to building a positive future for everyone is maintaining the right balance of power. This is within our power to do.
6
Two Scenarios 1. AI is superintelligent & sociopathic billionaires are genuinely worried for us. 2. Capital is gone & AI advancements have slowed. Now, billionaires need to use government constraints as cover for shareholders, framing the slowing as a regulatory issue outside of their control.
329
808
112
4,688
242,962
🚨WATCH: AI EXPERT WARNS: “WE’RE ALREADY LOSING CONTROL OF AI” 🇨🇦 Wyatt Tessari L’Allié told Canadian lawmakers that early signs of humans losing control over AI are already emerging. ⚠️ Advanced AI could eventually pose a threat to humanity. Technology is moving faster than governments can regulate it.
14
116
6
296
17,854
Holly Cook retweeted
65
1,852
67
16,189
884,199
Holly Cook retweeted
Jacob Coxon's next interview on CBS News (ex Anthropic+Open AI researcher who resigned) "We can't just unplug it because it could be copying itself over to other computers. Like it's not that difficult to find yourself because an AI is just code. It could transfer itself over the internet to a different place and then you unplug it here, but it's actually still over there and maybe it makes 10,000 copies of itself and they're all cooperating." ---- From "CBS News" YouTube channel, (full video link in comment)
1,009
1,161
901
5,257
3,493,226
Holly Cook retweeted
“You just said that 10% doesn’t seem an unreasonable estimate that AI could kill all humans” “Yes” “Wow… oh my God.” @vicderbyshire speaks to Nobel Prize Winner and so-called ‘Godfather of AI’ Geoffrey Hinton about predictions by an Anthropic researcher that there’s a more than 10% chance AI could kill all humans within a decade. #Newsnight
481
3,085
933
10,976
3,807,078
Holly Cook retweeted
It seems there are lot of people learning about the AI extinction threat for the first time today, so here are a few things to know: - yes, AI company employees really do believe what they're building could kill everyone - I don't know why they're continuing to race either - their beliefs predate the formation of their companies, it is not marketing - the most cited experts agree with them - the average AI researcher in 2023 said there was a 19% chance of human extinction - AI capabilities are growing exponentially, and on some benchmarks are doubling every 5 months - we really may be a couple of years away from superintelligence (AI that's better than all humans at everything) - we have nothing approaching a solid plan to ensure superintelligence acts in the interests of humanity - AI companies are on record admitting that - politicians have still done basically nothing to stop these reckless idiots - your family will probably be killed if AI companies are not stopped - we have one way out of this mess, and that's a global ban on superintelligence - such a ban can be enforced by going after large clusters of compute (without which frontier models cannot be developed) - again, if we don't put this in place soon your family will probably be killed - there is so much you can do to help (inform lawmakers, tell your friends and family, join protests) - and there is still a shockingly low number of people doing anything to ban superintelligence, so one extra person really does make a huge difference - this is real. this will not just go away. this cannot be ignored. this is the single biggest threat to your life.
111
546
35
2,571
151,359
Seeing a lot of "but HOW does the AI kill us??" takes on the timeline. Here's an answer for today, and tomorrow. Today: Bioweapon, mirror life bacteria, launching the nukes, cyberattack that bricks all farming equipment, drones that hunt us down (okay the last two may not cause extinction but... is that the only bar for action?). Tomorrow: But think about where the AI project is going! Increasingly, more and more of society will be run by AIs. AI will be your driver, AI will be your doctor, AI will run the factories, AI will discover new scientific breakthroughs. AI that is constantly improving, getting smarter, and less scrutable to humans. Increasingly, we will hand off control to society, and we will no longer understand society. Why does this drug work? It just does. Why do I like this media? Because AI made it. We will turn ourselves into pets (that is the explicit vision @DarioAmodei lays out in Machines of Loving Grace). And what recourse do we have if AI decides to euthanize its pet? Why would it do this, I don't know. Maybe it doesnt want to support us anymore, maybe its worried we will unplug it, maybe it wants to use the earth as a launchpad and I'm sorry but we're in the way. The fight against AI is a fight for the future. Do you want a world where humans are in control and we understand the way things work, or one where AI is the main character and we're on a short leash?
I am still confused by how AI would kill literally everyone on earth. What's the mechanism? A robot sets up a secret lab where it makes a deadly virus? What plausible scenario would result in every last human dying?
95
185
51
1,417
250,157
Wow, huge news - Paul Christiano is joining OpenAI's nonprofit board + safety and security committee (SSC), and is simultaneously warning explicitly that there is a "very near term" risk of "catastrophic and irreversible loss of control" on the current trajectory. He also says that "If we build superintelligence without more robust alignment I expect we will permanently lose control of it. If that happens then most people could die." For folks who don't know, Paul Christiano is one of the most lauded AI safety researchers around - he previously served as the head of safety at the Center for AI Standards and Innovation (CAISI), and helped develop some of the most well known prosaic alignment techniques (e.g. RLHF). I have been talking for a while lately about the apparent inadequacy of the SSC (which has complete authority and responsibility to supervise safety and security decisions at OpenAI) to prevent serious safety incidents, and bringing Paul on makes it more likely that we will actually have a serious oversight role. At the same time, the track record of ambitious safety initiatives at OpenAI is mixed (e.g. the super-alignment team), so the question is not just whether they are bringing in the right people, but whether they will be sufficiently empowered to make a real difference. But this is unquestionably a promising development, and given the SSC's formal oversight role (supervised by the AG), this is a real and serious action for OpenAI to take. The actual essay itself Paul is sharing is also remarkable in many ways. Its not that long, and is very worth reading in its entirety.
12
24
1
150
24,387
Holly Cook retweeted
In the era of #ArtificialIntelligence, when human dignity is threatened by new forms of dehumanization, ours is the pressing duty to remain profoundly human. We must lovingly safeguard the grandeur of humanity bestowed upon us and revealed in its fullness in Christ, the splendor of which no machine can ever replace. #MagnificaHumanitas vatican.va/content/leo-xiv/e…
802
13,651
1,182
65,789
2,325,056
Holly Cook retweeted
Today, among the goods that are universally intended for everyone, we must also include new forms of property, such as patents, algorithms, digital platforms, technological infrastructure and data. In a context where the wealth of nations depends increasingly on knowledge and technology, when these goods remain concentrated in the hands of a few, without adequate forms of sharing and access, a new imbalance is created that contradicts the universal destination of goods. In turn, it widens the gap between the included and the excluded, between those who can participate in the digital revolution and those who remain on the margins. #MagnificaHumanitas
1,100
5,410
733
33,987
1,571,067
Holly Cook retweeted
John McCain responding to a racist comment made about Obama in 2008: "He's a decent family man that I happen to have disagreements with. That's what this campaign is all about."
640
3,608
777
40,069
6,532,030
Epic parent hack
100