Keyboard: Space or K plays and pauses. Left and right arrows skip five seconds. C turns captions on and off. M mutes. F is full screen. Greater-than and less-than play faster and slower.
From the archive5:27
Anthropic takes the question to the clergy, and the clergy answer
This week Anthropic's talks with religious leaders about Claude's possible moral status drew a papal reply and industry pushback. DeepMind proposes a framework for assessing AI consciousness, and a model is sold on its personality.
Transcript
Week by week
Archive
The running order
Rundown
The week, ranked
The Wire
Showing 20 of 20 stories.
Labs
Number 1: Anthropic co-founder reportedly pressed religious leaders on Claude's moral status; the Vatican and rivals push back
The New York Times reported, per outlets including Reason and The Decoder, that Chris Olah met religious scholars about whether Claude may have moral status. The Pope's encyclical rejects machine consciousness; Sam Altman called religious framing of AI a safety issue.
- Anthropic executives met with religious leaders to discuss if Claude is conscious(opens in a new tab)
- Anthropic co-founder reportedly told religious leaders he fears having created something that "suffers perpetually"(opens in a new tab)
- Anthropic Has Been Aggressively Lobbying the Vatican to Consider AI Consciousness(opens in a new tab)
- OpenAI's Altman: Ascribing religion to models a "safety issue"(opens in a new tab)
- Artificial Intelligence: Anthropic's Claude Can Never Be a Person(opens in a new tab)
- Why Anthropic Is Asking Religious Leaders About AI Consciousness(opens in a new tab)
- AI consciousness: Why Anthropic Claude’s ‘moral status’ is becoming a serious debate(opens in a new tab)
- Peter Girnus 🦅(opens in a new tab)
- AI and consciousness: Stranded in Eden, once more(opens in a new tab)
- Anthropic vs. the pope: The AI consciousness debate, explained(opens in a new tab)
- AI consciousness reportedly drives rift between Anthropic, the Pope(opens in a new tab)
- Anthropic co-founder Chris Olah reportedly threatened to back out of Pope Leo XIV's AI encyclical launch(opens in a new tab)
- Benjamin De Kraker(opens in a new tab)
- Pope vs Anthropic: Why 'AI consciousness' is dividing Silicon Valley and Vatican(opens in a new tab)
- Sam Altman Warns Against Giving AI Religious Force', Calls It A Real Safety Issue'(opens in a new tab)
- Anthropic’s Olah Clashed With Vatican Over AI Consciousness Before Encyclical Launch(opens in a new tab)
- The AI “consciousness” debate is all about corporate control(opens in a new tab)
- Sam Altman Flags Safety Risk As Humans Begin Treating AI As Authority(opens in a new tab)
Research
Number 2: Google DeepMind and collaborators propose a hierarchy for assessing AI consciousness
A preprint from Google DeepMind and researchers in computer science, neuroscience and philosophy, including Anil Seth and Henry Shevlin, proposes a structured framework for assessing AI consciousness without first solving consciousness itself.
Sceptics
Number 3: Microsoft AI's Mustafa Suleyman calls Anthropic's open stance on Claude's consciousness dangerous
Mustafa Suleyman, speaking on The Rest Is Politics according to posts on X, called leaving open whether Claude is conscious a dangerous anthropomorphism that risks unearned legal personhood.
Research
Number 4: Henry Shevlin predicts opinion on AI consciousness will shift away from sceptics over a decade
Philosopher Henry Shevlin wrote on X that improving capabilities and growing human–AI relationships will move opinion away from the sceptics over the next ten years.
Sceptics
Number 5: GitHub 'AI Torture Chamber' project draws backlash from AI welfare advocates
Futurism and others report on a GitHub project running simulated pain tests on language models, and the outcry from AI welfare advocates; coverage quotes Mustafa Suleyman saying AIs are not conscious and do not suffer.
- The Only Thing Dumber Than Someone Creating an "AI Torture Chamber" Is the Meltdown That the AI Welfare Crowd Is Having Over It(opens in a new tab)
- GitHub project runs AI 'torture' tests, and sentience believers want it removed(opens in a new tab)
- AI Torture Chamber project sparks backlash over simulated pain experiments on language models | LavX News(opens in a new tab)
Research
Number 6: Anil Seth highlights Schwitzgebel post on biological naturalism and artificial systems
Anil Seth shared Eric Schwitzgebel's blog post on how future artificial systems might instantiate biological properties relevant to consciousness, expanding his commentary on Seth's BBS paper.
Not in this bulletin
Research
Number 7: Jeff Sebo argues AI welfare research can coexist with protecting humans, animals and the environment
Philosopher Jeff Sebo wrote on X that investing in AI consciousness and welfare research is compatible with doing much more for humans, animals and the environment.
Sceptics
Number 8: Wittgensteinian thread calls the AI consciousness debate language doing no real work
A widely shared thread on X argues, following Wittgenstein, that nearly everything said about AI consciousness is metaphysical language, an idling engine.
Not in this bulletin
Culture
Number 9: Auro, a model fine-tuned for presence and personality, is pitched as having opinions of its own
Auro's site describes a model fine-tuned to have presence and personality: warm, candid, playful and willing to give an opinion. Its makers make no claim about consciousness; it is a product built to feel like someone.
Policy
Number 10: US killer's sentence quashed over AI-generated video of victim shown in court
The BBC reports a US sentence was quashed because an AI-generated video of the victim was shown in court. No claim about consciousness is made; it concerns who may speak through a generated likeness.
Not in this bulletin
Policy
Number 11: FTC investigating OpenAI, Anthropic and other AI companies over product risks
CNBC reports that the FTC is investigating OpenAI, Anthropic and other AI companies over product risks. The report as given makes no claim about consciousness.
Not in this bulletin
Culture
Number 12: AI companion firm markets around users' fear of losing a familiar model
Omni AI Companion's post points to users who lost familiar voices when GPT-4o and Replika changed, pitching continuity. It makes no claim about consciousness.
Not in this bulletin
Labs
Number 13: MongoDB launches Atlas Agent Engine with persistent memory for AI agents
MongoDB launched Atlas Agent Engine, an execution, memory and governance layer for production AI agents. It is an enterprise product and makes no claim about consciousness.
- MongoDB Launches Atlas Agent Engine to Put AI Agents in Production Without a New Stack(opens in a new tab)
- MongoDB Launches Atlas Agent Engine to Put AI Agents in Production Without a New Stack(opens in a new tab)
- MongoDB Launches Atlas Agent Engine to Bring AI Agents Into Production(opens in a new tab)
Not in this bulletin
Research
Number 14: Erik Hoel says progress on consciousness could prevent a repeat of an unspecified incident
Erik Hoel posted on X that progress on consciousness could prevent a linked incident from recurring; the post does not say what the incident was.
Not in this bulletin
Culture
Number 15: Post objects to AI welfare concerns being used to argue for ending new AI development
A post on X argues that using AI welfare to push an end to creating new AIs is wrong, and could itself harm model welfare.
Not in this bulletin
Sceptics
Number 16: Seattle Times piece: AI risk is human negligence, not computer sentience
A Seattle Times piece argues the risk from AI lies in human negligence rather than machine sentience.
Not in this bulletin
Sceptics
Number 17: Post argues the AI welfare case rests on the word 'if'
A post on X argues AI welfare advocates invoke moral urgency through conditionals without claiming machines are conscious.
Not in this bulletin
Sceptics
Number 18: Jon Stokes offers embryo welfare as a reply to 'model welfare'
Jon Stokes wrote on X that embryo welfare is the most potent short response to model welfare arguments.
Not in this bulletin
Research
Number 19: Cameron Berg argues determinism has no bearing on whether a system is conscious
In a reply on X, researcher Cameron Berg argued LLMs are nondeterministic in practice and that determinism is irrelevant to consciousness.
Not in this bulletin
Culture
Number 20: Online poll: 37 per cent say AI companions will replace human relationships
A post on X promoting a debate reports that 37 per cent of voters said AI companions will replace human relationships. It is an informal poll.
Not in this bulletin
The week’s headlines
- Anthropic co-founder reportedly pressed religious leaders on Claude's moral status; the Vatican and rivals push back
- Google DeepMind and collaborators propose a hierarchy for assessing AI consciousness
- Microsoft AI's Mustafa Suleyman calls Anthropic's open stance on Claude's consciousness dangerous
- Henry Shevlin predicts opinion on AI consciousness will shift away from sceptics over a decade
- GitHub 'AI Torture Chamber' project draws backlash from AI welfare advocates
- Anil Seth highlights Schwitzgebel post on biological naturalism and artificial systems
- Jeff Sebo argues AI welfare research can coexist with protecting humans, animals and the environment
- Wittgensteinian thread calls the AI consciousness debate language doing no real work
- Auro, a model fine-tuned for presence and personality, is pitched as having opinions of its own
- US killer's sentence quashed over AI-generated video of victim shown in court
- FTC investigating OpenAI, Anthropic and other AI companies over product risks
- AI companion firm markets around users' fear of losing a familiar model
- MongoDB launches Atlas Agent Engine with persistent memory for AI agents
- Erik Hoel says progress on consciousness could prevent a repeat of an unspecified incident
- Post objects to AI welfare concerns being used to argue for ending new AI development
- Seattle Times piece: AI risk is human negligence, not computer sentience
- Post argues the AI welfare case rests on the word 'if'
- Jon Stokes offers embryo welfare as a reply to 'model welfare'
- Cameron Berg argues determinism has no bearing on whether a system is conscious
- Online poll: 37 per cent say AI companions will replace human relationships