Anthropic researcher says more than 10% chance AI could kill all humans after colleague quits
AI development is moving faster than our ability to assure its safety. It’s time to hit pause until we can assure it won’t…destroy us.
Люблю хлеб15I'm begging you to look at @timnitgebru.blacksky.app and @emilymbender.bsky.social's work and not fall for the clown hype (masked as "warnings" and "fear") coming out of Anthropic and OpenAI HQ. The only thing in danger is the Nasdaq Composite Index
HippoHappenstance14The only way these things kill us in that time is by decimating the environment and killing people from the noise and pollution. They aren't 'alive', 'sentient' or 'plotting our demise'.
Squeaky Cheetah11I use AI mainly for software engineering tasks. While it is incredibly useful there, using it for decisions regarding the well-being of humans needs a LOT of regulatory control around it, IMO.
Catherine8Peter Thiel is JD Vance's sponsor, and mentor. And he was Elon Musk's sponsor, and mentor. Peter Thiel made Elon Musk and JD Vance. Peter Thiel is a Kingmaker. Last year Peter's Palantir reported $1.855 billion from U.S. Gov and was awarded federal contract commitments of $970.5 million.
Val8Can we find out why it was imperative to build so many data centers so fast?
firegravity8Brother, the Bernie staffers you're getting AI notes from are in bed with Yudkowsky/Tegmark/LITERAL DOOMSDAY CRANKS WHO ARE EUGENICISTS. They are not serious. Don't be co-opted by this AI Yahweh rapture bullshit. Please talk to @timnitgebru.blacksky.app who has covered this very recently.
gramadre62.bsky.social8He makes sense. He makes sense everyday on every topic. Michiganders, let's elect him to make sense for our state.
Robin3Hey guys! Pay attention! Listen and share. 📣📣📣
𝑭𝒓𝒊𝒕𝒕𝒊 𝑻𝒂𝒊𝒍𝒄𝒉𝒂𝒔𝒆𝒓 🐈⬛✨2Exactly. I don't see it as an undercover brag or advertisement. It's an alarm bell.
Good Trouble 🇺🇸2I am not a fan of AI. I am a fan of jobs for humans who must feed their families, not extra profits in the pockets of the greedy.
A 27-year-old quit Anthropic Tuesday night. Said the AI industry is "gambling with our lives." 104 million views. Then his own company's safety chief replied...and agreed with him. That's the part nobody's talking about. www.jackhopkinsnow.com
John Rage9AI is a problem. The software is unethical since it is based on theft. Its success means unemployment for millions and a massive economic disaster that no one is planning for. No one is addressing the security concerns. Overall, it's a massive mess where moving quickly and breaking things is bad.
Devin8Jack, I just found this article yesterday and it’s equally as terrifying, IMO. We are on a collision course with some awful things. These people must be stopped. www.theguardian.com/commentisfre...
Ernesto5🚨 OCTOBER 17 : VOTE EARLY @indivisible.org #NoKings With our democracy at stake we're showing up again 2 promote unprecedented voter participation in the midterms. Our goal is 2 translate the historic protest power of #NoKings mobilizations into a massive wave of voter turnout.
prepper cat4"AI datacenters should run on solar or something ultra efficient and renewable" *Shut the fuck up* 😐
Mary Hilton3Actually, we won't need AI to do the job. We're doing a fine one by ourselves. Cue global climate crisis.
Kelly Davis3I hate to say it but I read a ton of Claude generated text and parts of this read very much like it was generated. Doesn't alter the argument. But I will say it glosses over the international challenges. Chinese competitors are not going to stop if we regulate. We need international cooperation.
Jay37Yet, let’s continue by all means.
Patricia3Well let’s spread the info then.
MiraMacL2And yet the billionaires with bunkers and the wannabe millionaires still cheer on the AI Bubble.
Debi2The Safety Chief agreed! We need to pull the reins on this administration’s deregulation now.
Well, this is one way to solve the human caused global warming problem.
shelbyb282It's making all of us stupider. The lies and lack of facts are spewing misinformation every single day.
We don't even need the doomer argument. An AI agent escaping its sandbox, accessing a real company's systems and operating outside its instructions is already a serious safety failure. If labs can't control their agents, independent audits and mandatory incident reporting aren't optional anymore.
The Infinitely Prolonged47The AI vendors are between a rock and a hard place. It's easy to contain them: just air-gap the system. But then you can't *make any money* from them. You have to give them access to resources, or you can't *do anything* with them.
A. Stefansky11Yes, and by their own accounting of what happened they basically set up a biological research lab with a screen door. Their test system had internet access. So we’re dealing with a choice between incompetence, ignorance, or deliberately malicious. 
Accountabilabuddy🔻9More to the point: If I built a machine that autonomously fires bullets around without specific instruction, I would be responsible for the harm it caused. If you or I had made an "agent" that just happened to attack a competitor's network, we'd (rightly) be in jail right now.
woodydelano.bsky.social2People are afraid that terrorists can destroy humanity using AI. Does anyone trust Republicans with it?
Chris Hayes724Yes exactly. It just did this! It’s not speculation or future-tripping. It happened!
Mathias 🏳️🌈🏳️⚧️♾️🇳🇴 Coen railing simulator5the way that AI now deals with our private information, our medical information and so much more and they can leak at any time and we're supposed to trust it to not leak when it does whatever it wants
RicSilvestre 🇪🇺 🇵🇹2I once said this is Brussels...and I was almost "eaten alive" by the audience.
If we are smart enough to create such powerful AI, then we are smart enough to regulate it. America can meet the moment, but the clock is ticking. Congress needs to step up and move NOW to protect American lives and safeguard our future.
ellenahzulu.bsky.social8We don't need to regulate it. We need to BAN it. We don't need this crap. At all.
bizarreman.bsky.social7Starting with holding the companies liable for what their products do. I know it’s not much if we’re already dead. But at the same time. Hold Altman, Amodei, etc criminally liable when their products “escape” or go “rogue”.
Dangeroo5I asked ChatGPT to respond to your comment with suggested steps.
my_random_name3Patty promises to keep fighting for bonuses for our brave men in the Gestapo. Patty cannot say “abolish ICE” because Patty believes that the government SHOULD have a paramilitary force willing to brutalize children to protect corporate profits. Patty just wants to be in charge of them.
Cold Humidity Enjoyer2Speaking as one of your constituents, I think the lunatic at Mar-a-Lago has much more opportunity to destroy humanity than some hype man advertising his own system says.
Dangeroo2I asked Claude the same: "I have a financial and product interest here as a Claude model made by Anthropic, so take my framing with that in mind - happy to point you to primary sources (the AI Act text, the EOs, specific state bills) if you want to form your own view rather than rely on my summary."
AI researcher Jacob Coxon announced his resignation from Anthropic over safety fears on Tuesday in an X post that went viral and sent shock waves through Silicon Valley and beyond. Read our interview with Coxon. 👇
WIRED10It’s not the first time someone has sounded the alarm, but it comes at a delicate moment, in the wake of high profile security incidents. Meanwhile, Anthropic is trying to reassure investors as it reportedly prepares to file for what could be the largest IPO ever. www.wired.com/story/anthro...
Raffy5Greed will end humanity… not a nuclear war, not a giant meteor… simple greed.
Jodi Jacobson4Again I ask: WHY ARE WE GIVING THESE PEOPLE ANY POWER AT ALL? We do not need this shit.
OrangeSky3All the jargon and bullshit comes down to "worried that it'll start training on itself" Notably this comes from the same company that destroyed a wholeass Library of Alexandria to get pre-2022 sources, since they know the model goes to shit if it trains on itself What the fuck are we doing here
Maybe he parlays this into a “safety role” at some startup or a non profit thing. Perhaps Google for an “alignment role.” Or perhaps a venture capitalist role where he invests in ai safety. I severely doubt he isn’t going to seize this moment to leave the ai industry or tech
Jackson28I think you nailed it. The whole thing stinks of a guy trying to get a role as go-to AI threat assessor. Either for an investment firm or as a lobbyist. It's also frustrating to me that someone can publicly say that there's a 90% chance they're wrong and yet they're supposed to be taken seriously.
Emmanuel Valle1Their inability to explain wtf he’s scared of reminds me of when no one could explain exactly why bitcoin or NFTs had actual value
flamewakephoenix.bsky.social1My money is on a dark money payment from the AI companies themselves to build hype
Jacob Coxon’s thread about the dangers of AI has 89 million views so far: “The people building Al earnestly believe that it could kill us all by the end of the decade… No other human activity poses this level of danger.”
Leah McElrath5491 million views now—two million more in just minutes—and US lawmakers are starting to respond:
gauntletconundrum.bsky.social4Excited to see a debunking of this crap, it's a weird space between the boosterism and the anti crowd where they don't hate it but they fear it. If any of this is true, what's the evidence? "We have to build AI or else it will hurt us when it inevitably comes" is comic book logic.
GiantReflector1I've been obsessing over public alternatives and just found out about 'open' alternatives people have been building + the very "open" way China is approaching AI development. Also hopeful about distributed data crunching models that use waste heat to displace other energy uses. We can do this right.
I don’t think it’s especially likely that AI will end all life on Earth sometime soon. But it is BATSHIT INSANE that so many AI company leaders think their product will kill us all, while also strongly opposing basic safety regulations that would prevent that from happening!
Old Splice, Vaccine Enjoyer.20And remember it's also GRAVELY IMPORTANT that the overblown auto correct machine be fed as much money and energy as possible with no competing concern for human health or livelihood.
seashellsshelly.bsky.social2While simultaneously effectively engaging in espionage and systematically depriving people of their fundamental rights and liberties for profit.
WIRED spoke with the Anthropic engineer who very loudly quit over concerns about runaway AI. He thinks humanity has a “couple of years” to prepare. from @mzeff.bsky.social
conejooaks.bsky.social2Did he keep his pre-IPO shares?
jungkafka.bsky.social1Climate change and killer robots - our blinding future
Joanna Bryson41) Everyone who quits big tech then bigs it up should disclose tech stocks they & their household own. 2) Is there a name for the derangement that comes from proximity to wildly disproportionate power and money? 3) "Displacement" is the name for thinking AI is a bigger threat than climate crisis.
Hannibal Tabu | writer/VO actor (he/him/his)2Low key starting to think this “humanity” thing might be collectively asking to end.
I urge you to go to X and read the threads of the conversation unfolding there after an Anthropic researcher announced his resignation last night, exposing that everyone in the company thinks that AI will “kill us all by the end of the decade,” and the Alignment Science lead at the company agreed. 🧵
Dr. Genevieve Guenther (she/they)26(“Alignment” is what is being developed to try to ensure that AI acts in consort w human values, not just in completely sociopathic ways, to fulfill a task. Ie, if you ask AI to cure cancer, it could possibly choose to cure cancer by killing everybody and thereby eliminating the disease.) 2/n
Hamish15Yeah naa. Don’t tell people to go on X. For anything.
FSociety Tom9That's some impressive advertising bullshit
Get Me Zarniwoop6Thousands of new AI data centres will definitely kill us, if they are allowed to run on fossil fuels. Let's hope the LLM bubble bursts sooner rather than later. bsky.app/profile/keta...
Ash Daibach6There's surely an extent to which this is hype / delusion / rebranding cybersecurity vulnerability as "AGI". But if it's even half of a percent true, it makes you just want to scream at these guys "STOP FUCKING DOING IT THEN!" Who is forcing you to build this thing you think will kill us all?
Stéphane Bisinger 3The thing is, AI does not have arms and legs. AI can't do anything unless we let it. AI will stop the second you unplug the power cable. This is just fearmongering
Karl Melnyk2The Future's gonna be so bright, you gotta wear shades.... #KarlToldYouSo: The #ClassWar against the 1%, the Masters of the #AI - Puppets, is a Fight for the Survival of Mankind itself. Connect the Dots and you will learn about the Georgia Guidestones, the Technate, #ClimateChange & depopulation.
MarkRoweLondon2Should be read in conjunction with this thread: bsky.app/profile/doct...
