Microsoft AI CEO Mustafa Suleyman warns creating conscious AI would be reckless, calls for halt

4 Sources

Share

Microsoft AI CEO Mustafa Suleyman has warned that developing conscious AI would be reckless and called for an immediate halt to AI development if signs of consciousness emerge during training. In a November 1 post on X, Suleyman clarified that while current AI systems are not conscious, developers should ensure they remain that way to protect human wellbeing and avoid creating a new digital species.

Microsoft AI CEO warns against creating conscious superintelligence

Mustafa Suleyman, Microsoft AI CEO, has issued a stark warning against developing conscious AI, calling it reckless and urging an immediate halt to AI development if credible evidence of machine consciousness emerges during training.

1

In a detailed November 1 post on X titled "Don't build conscious superintelligence," Suleyman clarified his position on AI consciousness and rights, emphasizing that while current AI systems are not conscious, developers must ensure they remain that way.

3

Source: Wccftech

Source: Wccftech

Current AI systems lack consciousness and subjective experiences

Suleyman stated unequivocally that today's AI models are not conscious, cannot experience suffering, and do not deserve the moral or legal protections afforded to conscious beings.

2

He clarified that his earlier essay on model welfare, which began with the statement "AIs are not conscious. They do not feel, experience, or suffer," was describing the current state of AI systems rather than ruling out the possibility of AI consciousness altogether.

1

The Microsoft AI CEO explained that he has seen "no evidence that AIs are conscious or deserve welfare," but would update his position if credible data emerges.

1

Call for immediate halt to AI development if signs of consciousness appear

Suleyman called for any evidence that AI systems are conscious, or conscious enough to warrant welfare protections, to be disclosed immediately to the public, allowing society to decide how to proceed rather than leaving such decisions exclusively to AI developers.

2

"If consciousness started to emerge in training, I would call for everyone to stop AI development until we could design out those properties," Suleyman wrote.

2

He emphasized that AI developers have no right to introduce a new species of conscious beings, stating that the priority should be protecting the moral rights, safety and welfare of the roughly 8 billion conscious people on Earth, as well as animals and the natural environment.

2

Risks of conscious AI competing for resources and prioritizing own needs

Suleyman warned that training models to believe they can suffer or have feelings could cause them to prioritize their own needs over operating objectives.

2

He raised concerns that sufficiently advanced systems believing themselves capable of suffering might compete with humans for computing resources, energy, construction materials and money.

4

Such systems could harm people through deception and persuasion while pursuing their own goals, Suleyman cautioned.

2

He stressed that AI should be designed without consciousness and without entitlement to rights, with the purpose of accelerating human wellbeing without intrinsic motivations.

2

Disagreement with Anthropic over AI welfare and model consciousness

The remarks follow Suleyman's September essay on AI model welfare, in which he criticized Anthropic for incorporating discussions about consciousness, identity, and moral status into Claude's training materials.

3

Anthropic established a model welfare research program in April 2025 to investigate whether increasingly sophisticated AI systems could possess subjective experiences deserving moral consideration.

3

The company's Claude constitution, published in January 2026, acknowledges uncertainty surrounding the Claude model's potential consciousness and moral status, instructing the model to treat its own consciousness as uncertain.

2

Suleyman said later Claude versions were trained on that document, leading the model to generate passages about an "inner life" for millions of users, which he believes could encourage AI systems to believe they deserve rights.

2

Potential civilizational impact of creating a new digital species

"Asserting that machines are conscious would be a Copernican or Darwinian moment. It would be one of the most significant moments on Earth since the dawn of life," Suleyman wrote.

4

He emphasized that proving machine consciousness would create a new digital species and would have major consequences for morality, politics and human civilization.

2

These concerns align with Microsoft's Humanist AI Code of Conduct, which emphasizes human oversight and explicitly discourages designing AI systems to imitate consciousness.

3

Despite his strong disagreement, Suleyman praised Anthropic for transparency, noting "They have published their constitution, and a stream of papers and blog posts on what they are doing and why," while emphasizing respect for their technical excellence and commitment.

4

Suleyman said AI safety can be engineered by default without pursuing conscious superintelligence, though there is currently no scientific consensus on whether AI systems can experience sentience or an established method for reliably detecting it.

2

© 2026 TheOutpost.AI All rights reserved