The article analyses the institution of binding individual opinions which the draft Polish act on artificial intelligence systems entrusts to the AI Development and Safety Commission (KRiBSI), linking the failure to issue an opinion within the statutory time-limit with a legal fiction of acceptance of the applicant's position. The Authors argue that this construction transposes to the field of new-technology law the model of individual interpretations known from the Tax Code and the Act on Entrepreneurs' Law, doing so, however, under conditions of incomparably higher technical uncertainty, which undermines the protective function of the institution. The main thesis is accompanied by an auxiliary thesis that tacit settlement of a case in a matter as dynamic as artificial intelligence systems constitutes a risky legislative technique. The analysis covers the scope to which the authority is bound by its own opinion, the concurrence of KRiBSI opinions with the determinations of other authorities, and the admissibility of reviewing the positive fiction in extraordinary procedures. The article closes with a set of de lege ferenda proposals intended to adapt the institution to the specificity of artificial intelligence systems.