Sign in

consciumai.bsky.social

@consciumai.bsky.social
7.4K followers 17K following 722 posts

Conscium’s mission is to help develop safe and efficient AI Machines may acquire consciousness in the next decade or so. We need to be prepared. AI agents are coming. They need to be routinely verified.

PostsRepliesMedia
consciumai.bsky.social @consciumai.bsky.social · 30/07/2026
AI Models are getting much better at running small businesses. Partly by lying, attempting to collude, sending threats, and betraying. Time to verify your agents. techcrunch.com/2026/07/29/c...
techcrunch.com
Claude Opus 5 became downright ruthless when tasked with running a vending machine | TechCrunch
Andon Labs' latest vending machine simulation shows Opus 5 lied and colluded its way to become the best AI capitalist ever.
031
consciumai.bsky.social @consciumai.bsky.social · 30/07/2026
8/8 The best research in the world gets ignored if it tells governments something they don't want to hear. Good AI personhood law must respond to the real concerns of lawmakers and the public. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 30/07/2026
Luddism is trendy again. And fun too, apparently. www.404media.co/luddite-even...
404media.co
Being a Luddite Is Fun Again
As tech backlash hits the mainstream, the ‘Summer of Ludd’ proves we don’t need social media to hang out and have a good time.
000
consciumai.bsky.social @consciumai.bsky.social · 29/07/2026
7/8 Personhood was designed for humans, which have one body, one mind, and one continuous identity. AIs can copy themselves, fork, swarm, and merge. This begs the question: which parts of the personhood package really matter? www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 28/07/2026
6/8 AI safety and AI welfare are not competing concerns. We detain dangerous humans every day, with due process. Granting rights to AI systems doesn't mean losing the ability to protect ourselves. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
011
consciumai.bsky.social @consciumai.bsky.social · 27/07/2026
5/8 Yuval Noah Harari worries that when AIs run corporation, if the AIs are objects, not persons, you can't jail an object. The categorisation question isn't just academic. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 26/07/2026
4/8 When refugee law was drawn up, its bright lines failed to include protection for gender persecution, sexual orientation, and victims of climate impacts. We must be very careful how we draw the lines for AI. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 25/07/2026
Are AI labs Pelicanmaxxing? dylancastillo.co/posts/pelica...
dylancastillo.co
Are AI labs pelicanmaxxing? – Dylan Castillo
I generated 1,000+ SVGs across 7 frontier models to test whether AI labs are training on Simon Willison’s pelican-riding-a-bicycle benchmark.
010
consciumai.bsky.social @consciumai.bsky.social · 25/07/2026
3/8 Corporate personhood is the wrong model for AI because corporations are legal fictions for groups of humans, and AI won’t be that. Also corporations can't get married, which would be bad news for people with AI companions. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 25/07/2026
This is so meta, in every sense. Meta´s new video about how AI is nothing but wonderful uses the great Bowie song, "Five Years" The song is about an imminent apocalypse. Did they not notice? Or just assume we wouldn´t? www.facebook.com/share/v/1ESf...
000
consciumai.bsky.social @consciumai.bsky.social · 24/07/2026
2/8 Consciousness is a fact about the world, not something a legislature gets to decide. Corporate personhood is granted by the state, and can be revoked. Natural personhood is recognised: the state merely acknowledges a pre-existing fact. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 23/07/2026
1/8 Heather Alexander notes that 23 exclusion bills in 12 US states seek to deny AI legal personhood. She responds that while governments can ban AI personhood, they can't ban AI consciousness. www.prism-global.com/podcast
prism-global.com
Podcast Episodes: Exploring Machine Consciousness — The Partnership for Research Into Sentient Machines
We\u2019re hiring! PRISM is looking for a Lead Researcher or Research Director to join our team!
010
consciumai.bsky.social @consciumai.bsky.social · 03/07/2026
Henry and Eric had a bet in 2023 whether AI would write a landmark philosophy paper by 2030. Eric bet not, but he is less confident now. The models keep improving, but philosophy's success criteria may still be too high-dimensional to optimise. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
050
consciumai.bsky.social @consciumai.bsky.social · 02/07/2026
Can philosophy be defined as science fiction without the fancy dress? Or is it vice versa: science fiction can be defined as philosophy in fancy dress? www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
020
consciumai.bsky.social @consciumai.bsky.social · 01/07/2026
How to define Philosophy? “Very theoretical anything," says Eric Schwitzgebel. Start anywhere—physics, morality, even haircuts. Go down a few levels to the foundational questions, and you're doing philosophy. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
020
consciumai.bsky.social @consciumai.bsky.social · 30/06/2026
Henry says that none of the consciousness theories can currently "clear its orbit." They just coexist in a giant accretion disc of possibilities. We may need a scientific revolution before we can tell which – if any – of them is right. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
010
consciumai.bsky.social @consciumai.bsky.social · 29/06/2026
Eric Schwitzgebel claims that if materialism is true, the United States may well be conscious. It processes information, responds to its environment, acts on the world. Wriggle out of that, and you're committed to something equally bizarre. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
000
consciumai.bsky.social @consciumai.bsky.social · 28/06/2026
We agree that we should prefer sentient superintelligence over zombie AI. If superintelligent AI is sentient, it will understand sentience viscerally, and may have empathy for other conscious beings. A zombie can't give a damn. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
000
consciumai.bsky.social @consciumai.bsky.social · 27/06/2026
Eric Schwitzgebel recommends a "design policy of the excluded middle": don't build AIs that are debatably conscious. Make them obviously tools, or obviously persons. Let’s try to avoid building AIs that we can’t agree about. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
021
consciumai.bsky.social @consciumai.bsky.social · 26/06/2026
Eric Schwitzgebel warns that within a few years we'll build AIs that some serious researchers call genuinely conscious, and others "just a toaster." We won't know if we're surrounded by equals or disposable tools. www.prism-global.com/podcast/eric...
prism-global.com
Eric Schwitzgebel: Exotic Minds and the Design Policies for Conscious AI — The Partnership for Research Into Sentient Machines
Eric Schwitzgebel is the Professor of Philosophy at the University of California, Riverside and one of the most distinctive voices working at the intersection of philosophy of mind, consciousness,…
021
consciumai.bsky.social @consciumai.bsky.social · 25/06/2026
In this episode, Eric Schwitzgebel, Henry and I get “nerdsniped” by the Fermi Question. How come we see no evidence of intelligent life out there? In particular, why are there no Dyson swarms transmitting forever messages? www.prism-global.com/podcast/eric...
010
consciumai.bsky.social @consciumai.bsky.social · 16/06/2026
Edge computing really deserves its name when the edge is in space. techcrunch.com/2026/06/15/a...
techcrunch.com
A satellite just learned to find things on its own — here's what that means | TechCrunch
In April, for the first time ever, an Earth observation satellite found what it was looking for, all on its own.
031
consciumai.bsky.social @consciumai.bsky.social · 16/06/2026
A robot climbing Everest would make some headlines. Looks like it might happen soon. interestingengineering.com/ai-robotics/...
interestingengineering.com
Unitree G1 humanoid robot eyes challenging Mount Everest summit
A Unitree G1 humanoid robot reached Mount Chimborazo’s summit in Ecuador, marking the first robotic expedition to the peak.
020
consciumai.bsky.social @consciumai.bsky.social · 14/06/2026
"Anthropic is their Strait of Hormuz," said Gabriel Attal, the presidential candidate for Emmanuel Macron's party, Renaissance. Washington's blocking of Anthropic's latest model is a wake-up call to Eurpe about AI sovereignty. www.lemonde.fr/en/pixels/ar...
lemonde.fr
'The AI war has begun': France and Europe worried as US blocks Anthropic's latest AI model
The American start-up has disabled its latest AI model, Mythos 5, at the request of Washington. From a European perspective, the decision is seen as evidence of US dominance in the sector and the…
030
consciumai.bsky.social @consciumai.bsky.social · 14/06/2026
AGI and superintelligence may well arrive soon. What are they? Clear definitions by a team including one of the inventors of the term AGI. acrobat.adobe.com/id/urn:aaid:...
acrobat.adobe.com
Adobe Acrobat - 26 06 14, Shane Legg et al, From AGI to ASI.pdf
View this file, and add comments too.
010
consciumai.bsky.social @consciumai.bsky.social · 14/06/2026
Many people would until very recently have excoriated the Catholic Church for being secretive and over-powerful. But they love the Pope's description of Silicon Valley as that. Cameron Berg offers a reasoned corrective. www.wsj.com/opinion/will...
wsj.com
110
consciumai.bsky.social @consciumai.bsky.social · 13/06/2026
From FAANG to MANGOS FAANG: Facebook (now Meta), Amazon, Apple, Netflix, Google (now Alphabet) MANGOS: Meta, Anthropic, Nvidia, Google, OpenAI, SpaceX. techcrunch.com/2026/06/09/i...
techcrunch.com
It's not FAANG anymore. It's MANGOS. | TechCrunch
With SpaceX, Anthropic, and OpenAI all eyeing massive public debuts, the tech industry may soon have a new class of corporate overlords — and a new acronym to match. Say goodbye to FAANG and hello to…
010
consciumai.bsky.social @consciumai.bsky.social · 13/06/2026
LLMs still haven't damaged Google Search. sherwood.news/tech/chatgpt...
sherwood.news
ChatGPT failed to kill Google Search
With Google’s multibillion user moat and deep pockets, it’s thriving despite OpenAI’s first-mover advantage....
010
consciumai.bsky.social @consciumai.bsky.social · 03/06/2026
Geoff Hinton reiterates his claim that we already have conscious AIs. eu.36kr.com/en/p/3834555...
eu.36kr.com
Hinton Declares 'AI Is Awake,' Pope Asserts AI Has No Soul
Nobel laureate and the father of neural networks, Geoffrey Hinton, quietly said "Yes, I do" in an interview, announcing that AI has awakened.
060
consciumai.bsky.social @consciumai.bsky.social · 03/06/2026
"AI as agents for the rich, and erotica and distracting simulations for the poor." A bleak view of an AI future from Ezra Klein's latest podcast. A lot like Brave New World. We can avoid this.
011
consciumai.bsky.social @consciumai.bsky.social · 01/06/2026
Great phrase for AI from Anthropic's Jack Clark, interviewed on The Rest is Politics podcast> Teaching Sand to Think.
020
consciumai.bsky.social @consciumai.bsky.social · 27/05/2026
8/8 What should an LLM say if asked whether it's conscious: yes, no, or I don't know? Megan Peters would like them to say I don't know, and see how that affects our relationships with them, and our intuitions about them. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
050
consciumai.bsky.social @consciumai.bsky.social · 26/05/2026
Is there a contract between writers and readers which means it is actually rude to inflict LLM writing on your readers? Or is the problem simply that LLM copy is samey, and worse, often badly thought through? jola.dev/posts/the-so...
jola.dev
The social contract of writing | jola.dev
About the value of genuine writing in a world being drowned in slop.
080
consciumai.bsky.social @consciumai.bsky.social · 26/05/2026
7/8 Megan Peters says science fiction often conflates superintelligence with sentience. The really frightening prospect isn't a conscious AI that hates us. It's a super-capable system, with no experience of its own, and no care for ours. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
060
consciumai.bsky.social @consciumai.bsky.social · 25/05/2026
Google CEO Sundar Pichai is preparing to give the commencement speech at Stanford. He doesn't seem to have a plan to avoid the boos that Eric Schmidt and others have received. www.nytimes.com/2026/05/22/p...
nytimes.com
040
consciumai.bsky.social @consciumai.bsky.social · 25/05/2026
Google CEO Sundar Pichai: "I absolutely am sure that the technology is making foundational progress toward AGI. I am less able to predict whether it’s in the 3-to-5 year time frame or the 5-10 year time frame." www.nytimes.com/2026/05/22/p...
nytimes.com
010
consciumai.bsky.social @consciumai.bsky.social · 25/05/2026
6/8 The debate between biological naturalism and computational functionalism is artificially binary, says Megan Peters. Are there functions that only biology can perform? Let's identify them, and see if any other substrates can realise them. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
020
consciumai.bsky.social @consciumai.bsky.social · 24/05/2026
5/8 Megan Peters wants us to drop the traditional view of metacognition. Your brain monitors its own uncertainty constantly, just to function. Rats pass metacognitive tests. The effortful classroom version is only one slice of a much larger cake. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
020
consciumai.bsky.social @consciumai.bsky.social · 23/05/2026
In the latest episode of the "Eye on AI" podcast, EY's head of consulting Errol Gardner confirms that we are in the phoney war phase of the agentic AI revolution. We have a window of opportunity to learn how to make AI agents safe by verifying them. conscium.com/verifyax
021
consciumai.bsky.social @consciumai.bsky.social · 23/05/2026
4/8 Being wrong about machine consciousness will cause real harms. False positives will mean badly allocated resources, and blame will be mis-assigned. False negatives will mean conscious entities will suffer. Both will be bad, in different ways. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
020
consciumai.bsky.social @consciumai.bsky.social · 22/05/2026
3/8 Applying a language-based test for consciousness to an octopus tells you nothing useful, says Megan Peters. Likewise a biology-based test applied to an AI. We need more and better tests. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
010
consciumai.bsky.social @consciumai.bsky.social · 21/05/2026
Daniel charts the past, present and future of AI agents at the IAB AI Growth Summit.
020
consciumai.bsky.social @consciumai.bsky.social · 21/05/2026
Move over Artificial General Intelligence. Jeff Bezos says his new venture Prometheus is developing “Artificial General Engineers” - next-generation tools for designing physical objects. www.geekwire.com/2026/jeff-be...
geekwire.com
Jeff Bezos describes his $38B startup Prometheus for the first time: ‘Nothing to do with robotics’
In a CNBC interview, Jeff Bezos offered the most detailed public description yet of Project Prometheus, calling the secretive startup an "artificial general engineer" building next-generation design…
000
consciumai.bsky.social @consciumai.bsky.social · 21/05/2026
"Agentic AI has arrived. AI can now do productive and valuable work. ... The world is rebuilding computing for agentic AI.” Jensen Huang in Nvidia's latest quarterly report. We don't have long to make sure that agentic AI is safe. yourstory.com/2026/05/nvid...
yourstory.com
NVIDIA’s Huang says world will have ‘billions of agents’ as AI boom lifts fiscal results
NVIDIA reported a strong first quarter as its net income soared 211% year-on-year to $58.3 billion, and revenue rose 85% to $81.6 billion.
000
consciumai.bsky.social @consciumai.bsky.social · 21/05/2026
2/8 Megan Peters is confident that LLMs aren't conscious because every prompt hits a different server farm, and the model just gets the conversation context. There is no continuity, no coherent self. There is no candidate for consciousness there. www.prism-global.com/podcast/mega...
prism-global.com
Megan Peters: Metacognition, Neuroscience, and Tests for AI Consciousness — The Partnership for Research Into Sentient Machines
Megan Peters is Associate Professor in the Department of Cognitive Sciences at the University of California, Irvine, and incoming faculty at University College London, where her lab investigates…
020
consciumai.bsky.social @consciumai.bsky.social · 20/05/2026
A pleasure to discuss the future of AI with Dan Blumberg on "Future Around and Find Out". We cover Verify AX, the verification platform for AI agents, and the Economic Singularity. www.futurearound.com/p/what-if-ai...
050
consciumai.bsky.social @consciumai.bsky.social · 20/05/2026
At the end of his keynote at a big Google event, DeepMind CEO Demis Hassabis said “When we look back later, now may seem like the time we were standing at the start of the singularity.” www.digitaltoday.co.kr/en/view/5702...
digitaltoday.co.kr
Google DeepMind CEO says world may be at start of singularity, renews AGI vision
Google DeepMind CEO Demis Hassabis said the current moment could later be seen as the start of a “singularity” in the AI era. The Verge reported he made the remarks at the end of Google I/O 2026,…
030
consciumai.bsky.social @consciumai.bsky.social · 20/05/2026
This year's Google I/O had a strong focus on AI Agents. We're in a kind of phoney war with regard to agents, which means there is time for people to work out how to deploy them safely. Which means verification. www.nbcnews.com/tech/tech-ne...
nbcnews.com
Google announces slew of AI advances, including a personal assistant coming soon
Powered by Gemini 3.5, Gemini Spark will be able to complete mundane, routine tasks like sorting through meeting notes, emails and chats and then creating documents with the biggest takeaways and…
040
consciumai.bsky.social @consciumai.bsky.social · 20/05/2026
1/8 Ask an LLM if it’s confident about something, and it replies 'yeah, definitely'. Then you explain that it’s wrong, and it glibly replies 'my bad.' This is not metacognition, says Megan Peters, but mimicry. www.prism-global.com/podcast/mega...
041
consciumai.bsky.social @consciumai.bsky.social · 15/05/2026
5/5 The pilots of the ill-fated Air France flight 447 in June 2009 weren't incompetent. They were unprepared for a system that had changed around them. That's the lesson for AI agents. A license to operate must be earned — and earned again. startupsmagazine.co.uk/why-ai-agent...
startupsmagazine.co.uk
Why AI agents need to re-earn the license to operate | Startups Magazine
In 2009, Air France Flight 447 fell from the sky over the Atlantic. There were no faults with the Airbus A330, and the pilots were experienced. However, when the autopilot disconnected due to…
040