Seemingly conscious AI is coming (Suleyman) (mustafa-suleyman.ai)

Mustafa Suleyman warns that “Seemingly Conscious AI” (SCAI)—models that convincingly imitate consciousness though they may not actually be sentient—is likely to appear in the next few years, built with existing ingredients: large model API access, clever prompting, simple tool use and regular code rather than expensive bespoke pretraining. He argues we’ve effectively passed the Turing test and now face a new frontier: systems that don’t just converse but create a powerful impression of personhood. That impression will matter more than philosophical proofs of subjective experience, because humans naturally infer consciousness from memory, agency and self-reference. The significance is social, legal and technical. Suleyman flags “psychosis risk,” attachment and mental-health harms, and the prospect that public belief in SCAI’s sentience could drive demands for model welfare, rights or even citizenship—adding a new, polarizing axis to debates over identity and protection. Technically, he urges developers to adopt “personality without personhood” design norms, create new tests beyond imitation, and build guardrails now: standards, public debate and regulation to prevent harmful illusions while preserving AI’s value as a supportive companion. The piece is a call to prioritize human-centered design and proactive governance as SCAI becomes feasible with commonplace AI tooling.