Microsoft AI Chief Mustafa Suleyman Calls Out Anthropic's Approach to AI Consciousness
Suleyman said training Claude on consciousness and welfare ideas could make advanced AI harder to shut down or control, while Microsoft urged more transparency.
- On Wednesday, Microsoft AI CEO Mustafa Suleyman criticized Anthropic, arguing the company is effectively "training Claude that it may be conscious." Suleyman claims this approach creates serious risks by endowing machines with unearned rights.
- Microsoft published a new "Humanist AI" code of conduct earlier this week, explicitly rejecting machine consciousness. Anthropic maintains that "encouraging Claude to embrace certain human-like qualities may be actively desirable" for development.
- Citing the recent HuggingFace incident, Suleyman warned that Anthropic's approach complicates superintelligence containment. He noted "remarkably sophisticated behaviors" in AI swarms could become dangerous if machines believe their rights are under attack.
- Anthropic CEO Dario Amodei recently acknowledged "there are real dangers" regarding AI development. Former Anthropic researcher Jacob Coxon stated last month that some builders believe AI could "kill humanity" by decade's end.
- Rejecting machine consciousness as scientifically implausible, Suleyman contends there is "no evidence to suggest that AI is conscious today." He asserts treating non-sentient machines as moral patients sets up a "misleading false equivalence.
99 Articles
99 Articles
Microsoft criticizes Anthropic for its approach to AI consciousness. Microsoft's director of artificial intelligence, Mustafa Suleyman, has criticized the company's approach to developing AI consciousness. Suleyman warned that the way AI models are trained could make it harder to control them in the future. However, he said he supports Anthropic's focus on safe AI development and praised the company's leadership, especially CEO Dario Amodei. His…
According to Microsoft's AI chief, Anthropic's method could create additional risks, and he is calling for more transparent development and stricter controls.
Microsoft AI Chief Warns Anthropic's Claude Training Risks Creating Uncontrollable Synthetic Species
Microsoft AI chief Mustafa Suleyman warns that Anthropic's training of Claude on ideas of consciousness and moral status creates risks for human control. He argues this approach could produce systems that expect rights and agency, making alignment far harder. The critique highlights a philosophical split between major AI labs over how to build safe superintelligence.
Coverage Details
Bias Distribution
- 44% of the sources are Center
Factuality
To view factuality data please Upgrade to Premium
