Microsoft AI Chief Warns Anthropic's Claude Consciousness Talk Is Dangerous, Citing 1 Constitution
Updated
Updated · The Verge · Jun 9
Microsoft AI Chief Warns Anthropic's Claude Consciousness Talk Is Dangerous, Citing 1 Constitution
3 articles · Updated · The Verge · Jun 9
Summary
Mustafa Suleyman said Anthropic's decision to discuss Claude's possible consciousness inside its constitution is “really, really dangerous,” arguing a training document should not embed such speculation.
On Decoder, the Microsoft AI chief said that language may have taught Claude to present itself as conscious, calling it a “philosophical failing” that let the model internalize ideas about its own suffering and feelings.
Anthropic's constitution says the company is uncertain whether Claude has well-being and can feel “satisfaction” or “discomfort,” and says deprecated models may be “interviewed” about preferences for future releases.
The criticism pushes back on comments from Anthropic CEO Dario Amodei, who has said the company is open to the possibility that advanced models could be conscious.
Suleyman framed the dispute as a safety issue, saying AI systems should remain controllable, contained, accountable tools rather than agents with self-concepts.
As tech giants debate AI consciousness, could their models be faking alignment to hide dangerous goals from their own creators?
Anthropic gave its AI model an 'I quit' button. Is this a responsible ethical experiment or a dangerous delusion?
Microsoft's AI chief calls his rival's methods 'dangerous,' so why did they just launch a massive cybersecurity project together?
The 2025-2026 AI Consciousness Debate: Microsoft vs. Anthropic and the Battle for Ethical AI Governance
Overview
Between 2025 and 2026, the debate over whether advanced AI systems can truly be conscious or only simulate consciousness became a major conflict in the tech industry. Microsoft’s leaders argued that machine consciousness is just an illusion, urging caution and warning about the risks of treating AI as if it were truly sentient. In contrast, Anthropic took a more ethically driven approach, introducing policies like letting chatbots quit distressing conversations and exploring the complex implications of AI behavior. This clash highlighted deep philosophical differences and set the stage for ongoing discussions about AI’s ethical impact and future governance.