Research On The Identification Of Autonomous Algorithm Collusion From The Perspective Of Anti-monopoly Law

In the anti-monopoly law,the regulation of monopoly agreements can start from the three elements of subject,behavior,and anti-competitive effect.In the face of algorithmic collusion,the corresponding provisions are not specific and detailed enough.Different from traditional monopoly agreements,operators use algorithms to collect and analyze operator data,observe and track operators business strategies,price information,etc.traditional form.When the algorithm acts as an auxiliary tool for the business decision-making of the operator,the collusion reached at this time is no different from the explicit collusion,and it can still be regulated by the anti-monopoly law.However,when the algorithm is combined with artificial intelligence technology,the autonomous algorithm can achieve the result of collusion without human intervention in the operation of the algorithm.This kind of autonomous collusion lacks the subjective intention of the operator and is based on the transparency of the data market.With higher stability,the traditional regulatory measures of anti-monopoly law are difficult to use.Specifically,in autonomous algorithmic collusion,the operator’s conspiracy intention is more concealed,and law enforcement agencies cannot determine whether the collaborative behavior achieved by the algorithm is the algorithm’s autonomous behavior or the operator’s "behind-the-scenes" agreement to facilitate intentional contact.At the same time,because the operator lacks or even cannot directly intervene in the operation process of the algorithm,the identification of the responsible subject of the autonomous algorithm will also be in trouble.Based on this,the identification dilemma of autonomous algorithm collusion should be solved one by one.First of all,given the high concealment and difficult identification of the collusion intention of the operators,the cooperative behavior presumption mechanism based on appearanceism can be adopted.Conspiracy intent,on the other hand,the suspicion of complicity is excluded by self-certification of algorithm users and designers to prove that they do not have the possibility of conspiracy intent.Secondly,the dilemma of the indistinguishability between the algorithm coincidence of the autonomous algorithm and the intentional connection of the operator can be solved through three aspects.Its own rules and logic operation and decision-making;the second is to establish a data evaluation mechanism to evaluate whether algorithm designers and users deliberately feed abnormal data during algorithm training and trial operation,or deliberately feed,filter,or mark certain data for use.Algorithms form a tendency to collude;the third is to establish a mechanism for compulsory disclosure of algorithms.By removing barriers to trade secrets,algorithms can be presented more transparently,reducing the work pressure of law enforcement agencies to monitor and collect data as evidence.Finally,return to the problem of the imputation of the collusive subject,establish the identification standard that the user is the general and the designer is the supplement,and divide the imputation subject of the autonomous algorithm colluding.By starting with the effect elements,behavior elements,and main elements of traditional monopoly agreements,an autonomous algorithmic conspiracy identification framework is established,to contribute to the smooth operation of the future market competition order.