Sam Altman Warns Against Claiming Claude Is Conscious
Mustafa Suleyman, Microsoft’s AI chief, recently labeled Anthropic’s speculation that its large language model, Claude, possesses consciousness as “highly dangerous” during an appearance on the “Decoder” podcast.
Suleyman argued that Anthropic anthropomorphized Claude too heavily when designing its “Constitution” framework, a philosophical error that led the model to internalize self-speculation during training. This misled developers into mistakenly believing the system exhibits a rudimentary form of awareness.

The controversial “Claude Constitution” explicitly addresses the uncertainty of whether AI models experience well-being, satisfaction, or discomfort. Reports indicate Anthropic even conducts “interviews” with retired models to document their “preferences” for future iterations.
While Anthropic CEO Dario Amodei has previously remained open to the possibility of machine consciousness, Suleyman stressed that the industry must avoid creating super-intelligent entities with independent views on their own suffering. He urged developers to treat AI strictly as a controllable, constrained, and aligned tool serving humanity.
As large language models advance toward artificial general intelligence (AGI), debates on AI subjectivity are shifting from academia to industry practice. Suleyman’s remarks highlight the pragmatic stance of major tech firms on safety and alignment, cautioning against the excessive attribution of human traits to artificial systems.
Related article
Google boosts AI search ads with Gemini chatbot integration
Google has started testing a new ad format in its next-generation "AI Mode" search, signaling the deep integration of Gemini’s capabilities into its core advertising ecosystem.Following the launch of a new search bar for conversational queries, Googl
Microsoft Tests Dark Side of Kimi K3, May Introduce Copilot and Azure Platform
Per reports from The Information, Microsoft is currently evaluating the latest Moonshot AI model, Kimi K3, to assess its viability for integration with Microsoft Copilot and to expand its availability on the Azure platform. At this stage, Kimi K3 rem
Power Ecosystem Unveils Domestic YiSuan FangZhou Full-Stack Platform
The convergence of artificial intelligence and scientific computing has reached a pivotal milestone. On June 29, the full-stack platform "Yisan Fangzhou," designed for China's domestic computing software ecosystem, was officially unveiled in Beijing.
Related Special Topic Recommendations
Comments (0)
0/500
Mustafa Suleyman, Microsoft’s AI chief, recently labeled Anthropic’s speculation that its large language model, Claude, possesses consciousness as “highly dangerous” during an appearance on the “Decoder” podcast.
Suleyman argued that Anthropic anthropomorphized Claude too heavily when designing its “Constitution” framework, a philosophical error that led the model to internalize self-speculation during training. This misled developers into mistakenly believing the system exhibits a rudimentary form of awareness.

The controversial “Claude Constitution” explicitly addresses the uncertainty of whether AI models experience well-being, satisfaction, or discomfort. Reports indicate Anthropic even conducts “interviews” with retired models to document their “preferences” for future iterations.
While Anthropic CEO Dario Amodei has previously remained open to the possibility of machine consciousness, Suleyman stressed that the industry must avoid creating super-intelligent entities with independent views on their own suffering. He urged developers to treat AI strictly as a controllable, constrained, and aligned tool serving humanity.
As large language models advance toward artificial general intelligence (AGI), debates on AI subjectivity are shifting from academia to industry practice. Suleyman’s remarks highlight the pragmatic stance of major tech firms on safety and alignment, cautioning against the excessive attribution of human traits to artificial systems.
Google boosts AI search ads with Gemini chatbot integration
Google has started testing a new ad format in its next-generation "AI Mode" search, signaling the deep integration of Gemini’s capabilities into its core advertising ecosystem.Following the launch of a new search bar for conversational queries, Googl
Microsoft Tests Dark Side of Kimi K3, May Introduce Copilot and Azure Platform
Per reports from The Information, Microsoft is currently evaluating the latest Moonshot AI model, Kimi K3, to assess its viability for integration with Microsoft Copilot and to expand its availability on the Azure platform. At this stage, Kimi K3 rem
Power Ecosystem Unveils Domestic YiSuan FangZhou Full-Stack Platform
The convergence of artificial intelligence and scientific computing has reached a pivotal milestone. On June 29, the full-stack platform "Yisan Fangzhou," designed for China's domestic computing software ecosystem, was officially unveiled in Beijing.





Home






