Mustafa Suleyman
Mustafa Suleyman, the CEO of Microsoft AI, has issued a warning over how rival artificial intelligence (AI) companies build and market AI, arguing that treating such tools like sentient human beings risks creating an uncontrollable “silicon species” that will end up competing directly with humanity. Appearing on BBC, Suleyman took direct aim at competitor Anthropic, describing its methodology as deeply misguided. He cautioned that giving AI agents human-like personas, personal goals and financial agency threatens to unleash systems that society cannot rein in. Suleyman insisted that practical safety mechanisms still exist to keep autonomous models in check While acknowledging that public anxiety regarding rapid computing breakthroughs is entirely warranted.
“We must not sleepwalk our way into a decision we later come to bitterly regret,” he wrote. He also said that Microsoft is developing AI to work with humans and AI should not resist shutting down. he openly criticised the lab for anthropomorphising software – a technique that projects human personality, emotional depth and independent identity onto lines of code While Suleyman praised Anthropic CEO Dario Amodei and his technical team as intellectually honest researchers. By dressing models up in human clothing, Suleyman argued, software labs risk unleashing systems that become impossible to manage. To prevent that outcome, he called for urgent transparency across training data, independent behavioral audits, and robust kill-switches and monitoring toolsets.
Suleyman’s comments came after he published a detailed essay earlier this week, questioning how Anthropic trains its signature model, Claude.