OPINION OF THE COURT
Bernard J. Fried, J.The issue in this case is whether it was unlawful to arrest the driver of an automobile on the basis of a false and erroneous computer-generated police alarm that the automobile had been stolen three months earlier, when it had actually been recovered three days after the theft but the alarm had never been canceled. For the reasons stated below the arrest, under the circumstances of this case, was unlawful and the evidence seized must be suppressed.
According to Police Officer William Corcoran who testified credibly at a suppression hearing, Richard Dorsey reported the theft of his 1978 Oldsmobile on July 15, 1980 to the 108th Precinct in Queens. At that time, the fact of *876the theft was entered in the police computer causing an “alarm” to be established for the stolen vehicle. Three days later the car was recovered and returned to Dorsey by police officers from the same precinct; however, for reasons that do not appear in the record, there was a failure to correct the information in the computer data bank to show that the automobile had been recovered. In plain language, the “alarm” was not canceled or rescinded.
Thereafter, on October 16, 1980, at approximately 2:25 a.m., almost three months after the recovery of the stolen vehicle, Police Officer Corcoran and his partner were on radio patrol in the 28th Precinct in Manhattan, driving a marked police vehicle equipped with a mobile field computer. According to Officer Corcoran, by entering a license plate number into this unit, he could determine if there was an “alarm” for a particular automobile, i.e., whether it was reported stolen. Officer Corcoran testified that, in his experience, almost 20% of such “alarms” are erroneous.1
While on patrol, the officers observed a 1978 Oldsmobile with Illinois license plates stopped in a bus stop at West 125th Street and 7th Avenue. Officer Corcoran entered the license plate number into the field computer and received notification that there was an “alarm” for the vehicle. By this time, however, the car had left the bus stop and the officers proceeded to follow it north on 7th Avenue. While following behind the car, Officer Corcoran radioed the license plate number to the police radio dispatcher and received confirmation of the field computer readout that the Oldsmobile was a stolen vehicle. The officers then put on a flashing red light and pulled the car over. Officer Corcoran approached the driver of the automobile, Gloria Stone, told her to step out of the car, and advised her that she was under arrest for possession of a stolen vehicle. When Ms. Stone responded that the car was not stolen, Officer Corcoran then used his portable radio and again radioed the license number to the dispatcher. He was again told that the 1978 Oldsmobile was a stolen car and that it had been stolen on July 15, 1980. Without any further inquiry, Officer Corcoran and the approximately six other *877police officers, who had responded to the arrest, informed Ms. Stone’s two passengers, the defendants Cecil Jones and James Peterson, that they were also under arrest and ordered them out of the automobile. At no time, to this point, did Officer Corcoran make further inquiry or request to see the vehicle registration.
Thereafter, under circumstances not here relevant, controlled substances were seized from Peterson.2 The three arrestees were taken to the precinct for processing and additional controlled substances were seized from Peterson, as well as from Jones.
At the precinct, following the search of her two companions, Ms. Stone’s pocketbook was searched and she produced the vehicle registration for the 1978 Oldsmobile and gave Officer Corcoran the telephone number of the owner, Richard Dorsey. Corcoran called Dorsey, who came to the precinct about one-half hour later and stated that although his car had been reported stolen to the 108th Precinct on July 15, 1980, it had been recovered and returned to him by that precinct three days later. After Dorsey provided vehicle identification for the 1978 Oldsmobile to Officer Corcoran, Corcoran personally ascertained that the “alarm” had never been canceled by the 108th Precinct and “personally canceled the alarm with the Communications Division and the Alarm Board.” Ms. Stone was then released; her two companions, however, were processed on the instant charges.
Analysis of the lawfulness of the police conduct begins with People v Lypka (36 NY2d 210), where the Court of Appeals held that law enforcement authorities may properly rely on and act upon a police radio bulletin, and that where such bulletin establishes probable cause, then an arrest and subsequent search is authorized. Such police conduct is countenanced because “the sender’s knowledge is imputed to the receiver and, when the receiver acts, he presumptively possesses the requisite probable cause” (supra, at p 213). However, not only must the bulletin on its *878face establish probable cause, but the sender must have possessed probable cause at the time the bulletin was dispatched since, as the Supreme Court held in Whiteley v Warden (401 US 560, 568), “an otherwise illegal arrest cannot be insulated from challenge by the decision of the instigating officer to rely on fellow officers to make the arrest.” Ordinarily, then, when the radio bulletin is. challenged, the People are required to call the fellow officer (sender) to prove that he possessed sufficient information to justify the arresting (receiving) officer’s action. (People v Havelka, 45 NY2d 636, 641.) This was not done in the instant case; however, there is no dispute that the automobile was recovered three days after it had been stolen and the recovery was approximately three months prior to Officer Corcoran’s reliance on the police alarm. Thus, this defect, if any, is not fatal (cf. People v Jenkins, 47 NY2d 722), and it is necessary to consider whether the arrests were valid.
Amazingly, in this computer age, there are no reported decisions in this State concerning the Fourth Amendment consequences of an erroneous computer-generated police alarm for a stolen automobile. Close, however, is People v Lemmons (49 AD2d 639, affd on other grounds 40 NY2d 505), where an automobile which defendant was driving was stopped by State troopers for a speeding violation. When stopped, Lemmons produced a Michigan driver’s license and a New York registration form. Following standard procedures, the officers radioed the police dispatcher for a computer check of the documents and were advised that Lemmons was wanted in Michigan on a fugitive warrant resulting from a weapons violation. Based on this computer-generated information, Lemmons was arrested as a fugitive from justice and removed from the vehicle. Thereafter, when the officer returned to Lemmons’ car to determine the identities of the three passengers, he observed, in plain view, the handle of a .45 caliber automatic pistol. A subsequent search uncovered additional weapons and all three passengers were placed under arrest. It was only after these events that it was ascertained that the Michigan fugitive warrant against Lemmons had been *879dismissed a few days before the traffic stop, and that apparently the computer entry had never been corrected, rescinded or canceled. On these facts the Appellate Division, relying on Lypka and Whiteley v Warden, held that the arrest of Lemmons, on the basis of the erroneous computer-generated information that there was an outstanding fugitive warrant, was invalid. The court did, however, sustain the seizure of the weapons under the “plain view” doctrine. (See Coolidge v New Hampshire, 403 US 443.)
The Appellate Division decision in Lemmons, although strikingly relevant, is not dispositive since the Court of Appeals in affirming the conviction relied on the “plain view” doctrine and in a footnote stated that “[i]n light of our resolution of the issue, we do not decide whether Lemmons’ arrest was, in fact, valid.” (40 NY2d 505, 508, n 3, supra.)3 The court then cited to People v Lypka (36 NY2d 210, supra) and People v La Pene (40 NY2d 210, 223-224), where an anonymous telephone tip, absent additional circumstances, did not justify a stop and frisk. This not altogether clear reference to Lypka and La Pene may suggest that the Lemmons erroneous police bulletin, being more reliable than the La Pene police bulletin which was based on an anonymous phone call, was sufficient to authorize limited police action, although not an actual arrest.
The only other pertinent New York decision is People v Lent (105 Misc 2d 831), which dealt with the arrest of a defendant based on a telephone report that there was in existence an arrest warrant when, in fact, the warrant had been vacated earlier in the day. Referring to Lypka, and pointing out that the delay in making notification that the warrant had been vacated was “relatively brief — being only a part of one day” (supra, at p 834), the court held that there was probable cause to arrest on the basis of the teletype, and “[t]he mere fact that unknown to [the arresting officers] the warrant had been vacated earlier that day” (supra, p 836) was of no ultimate consequence.
*880Lacking decisions on point in New York it is instructive to turn to decisions from other jurisdictions. One such case is Carter v State (18 Md App 150), where the Court of Special Appeals of Maryland treated a factual situation substantially similar to that here. In Carter, the defendant was a passenger in a car that had been stopped for a traffic violation. After the car was pulled over, the police officer determined from the radio dispatcher that the license plates had been stolen three months earlier. Based on this information, which turned out to be erroneous, the plates having been recovered 10 minutes after they were reported stolen, the occupants of the car were arrested. It developed at the hearing that the police records were never corrected nor was the alarm ever canceled or rescinded. On these facts, the court concluded that there was lacking probable cause to sustain the arrest, cogently stating that “[i]t is one thing to permit an arrest to be made on information residing in the police files; it is another to sanction an arrest the probable cause for which is based on police information which is not true” (supra, at pp 154-155). In another case, United States v Mackey (387 F Supp 1121, 1125, n 9) the defendant was arrested solely on the basis of computer-generated information that he was wanted in another State for a probation violation, when in fact, the warrant had been satisfied five months before the events in question. Suppressing the evidence seized from defendant, the District Court concluded “that a computer inaccuracy of this nature and duration, even if unintended, amounted to a capricious disregard for the rights of defendant as a citizen of the United States” and went on to hold that there had been a deprivation of due process of law, finding it unnecessary, however, to consider the Fourth Amendment arguments. Finally, in People v Decuir (84 Ill App 3d 531) where an arrest, based on a 10-month-old warrant was invalidated because the warrant had, in fact, been quashed and recalled 16 days after it had been issued, almost Wz months earlier. The court in Decuir applying a straight Whiteley analysis stated: “[s]ince the warrant issued in the case at bar had been quashed and since there does not appear to be any probable cause aside from the warrant for *881the arrest of the defendant, the reasoning in Whiteley appears to be controlling” (supra, at p 533).
Seeming contrary cases are Patterson v United States (301 A2d 67 [DC App]), Childress v United States (381 A2d 614 [DC App]), State v Cross (164 NJ Super 368), and Commonwealth v Riley (425 A2d 813 [Pa Super Ct]), where arrests and incidental searches and seizures, based on erroneous computer or teletype information, were sustained. In each of these cases, however, there were additional circumstances which, taken together, justified the police conduct, the absence of which, on analysis, would likely have resulted in suppression being granted.
In Patterson, an automobile, bearing dealer license tags, was reported stolen and recovered approximately one week later during the early morning hours of February 4, 1972. Although the police teletype and computer units were notified that the automobile and tags were recovered, the police records were not corrected. Some 15 hours later, on February 4, a police officer, assigned to stolen car duty wrote down, on a paper prepared for patrol duty, that the automobile was still stolen, which information was taken off the erroneous police teletype, although the officer testified that he had never before known such teletype information to be incorrect. Possessing this erroneous information, he observed a different automobile with the same license plates that had been on the previously reported stolen car. Following this observation, the radio dispatcher was requested to run a check, which was done with the police teletype and the National Crime Information Center computer, and it was ascertained that there was an outstanding alarm for the vehicle. The arrest of Patterson ensued. The District of Columbia Court of Appeals, pointing out that the arresting officer’s experience was that the police teletype had never been incorrect, found probable cause, notwithstanding the fact that the teletype was erroneous. Implicit in this decision, as was made explicit in a subsequent decision of that court, Childress v United States (381 A2d 614, 617, supra [DC App]), is the significance of the brief interval between recovery of the stolen automobile and the arrest of the defendant. As held in Childress (supra, at pp 617-618), which involved a four-day delay *882(including a weekend) between the date that the outstanding warrants were satisfied although the computer records were not corrected, the “combination of reasonable administrative delay and reasonable police reliance on the misinformation produced by such a delay” did not warrant suppression. The court noted (supra, pp 617-618, n 3) that the four-day delay did “not rise to the level of police administratative negligence fatal to the government in Mackey”. It seems clear, therefore, that Patterson and Childress, read together, suggest that, in a situation such as here, where there was an inordinate and unexplained delay in correcting the records, coupled with a police officer who was personally aware of the relatively high degree of inaccuracy in “alarms” for stolen cars and failed to make additional inquiry following protestation that the car was not stolen, the District of Columbia Court of Appeals would invalidate the police conduct.
The Cross case (164 NJ Super 368, supra), does not point to a contrary result here. In Cross, the defendant, who was a passenger in a car that was stopped by the New Jersey State Police for speeding, gave the police officer the vehicle’s registration. Thereafter, the officer radioed for a computer check on the car and was advised that the car had been reported stolen by the Camden Police Department. The officer, while awaiting assistance was notified, by radio, that the Camden Police Department had been contacted and had confirmed that the car was still stolen. The defendant Cross and the driver were then arrested, although Cross insisted that the car was his and that it had been returned to him after he had reported it stolen. Indeed, as was later determined, the car which had been stolen a little over a month earlier, had been returned to Cross;' the Camden police simply “forgot” to cancel the computer alarm. This insistence on the part of Cross caused the officer to seek further personal identification of Cross and the driver. Failing to obtain any from Cross, and believing that the car was stolen, the officer looked in the glove compartment from where various pills and tablets were seized. In reaching its decision, the Cross court cited Carter v State (18 Md App 150, supra), with a “But see” notation, and then pointed out the reasonableness and *883good faith of the police conduct. Of course, in Cross, the arresting officer did not, contrary to the case here, ignore the claim that the car was not stolen but made further inquiry. Under the circumstances, such police conduct was not unreasonable. Moreover, direct and further inquiry was made with the sending police department to determine the viability of the alarm. This attempt to verify the officer’s belief that there was an outstanding warrant permitted the court to hold that the officer acted in good faith; all that could reasonably be expected was done, unlike the situation here where Police Officer Corcoran simply rechecked the original alarm. Nor does Commonwealth v Riley (425 A2d 813, supra [Pa Super Ct]), involving an arrest based on a warrant that had been vacated four days earlier require a different result. Riley agreed with the Childress case and pointed out that it would not invalidate an arrest based on “a computerized information only four days out of date” which the officer “had no reason to believe was incorrect” (425 A2d, at p 816). This reliance on the shortness of the interval coupled with the lack of evidence that the officer knew of inaccuracies in such computer-generated information, distinguishes Riley from the instant case.
This difficult problem of inaccurate or outdated computerized or similar police records has been addressed by Professor La Fave, who states that “[t]he point is not that probable cause is lacking because it turned out that the Tacts’ upon which the officer acted were actually not true, for quite clearly information sufficient to estabish probable cause is not defeated by an after-the-fact showing that this information was false, any more than information insufficient to show probable cause can be found adequate on the basis of an after-the-fact showing that in fact the conclusory allegations were correct. Rather, the point is that the police may not rely upon incorrect or incomplete information when they are at fault in permitting the records to remain uncorrected.” (La Fave, Search and Seizure, § 3.5, p 636; emphasis in original.) To permit otherwise would be to permit the police to profit from their own lack of responsibility in developing and adhering to procedures “de*884signed to minimize the risk of error”. (Cf. Baker v McCollan, 443 US 137, 156 [dissenting opn per Stevens, J.].)
Certainly the risk of error and resultant police conduct must be weighed against the societal need for police officers to use, and be able to rely on, the most modern and sophisticated equipment. And especially important, in an era when “[t]he practice of making a radio check with a centralized data bank is now a routine policy, followed * * * in literally hundreds of thousands of cases per day nationwide” (Baker v McCollan, supra, at p 155), is that police officers receive transmitted information that is reliable to a substantial certainty. However, as illustrated by this case, such reliability does not conform to actuality. Here, Officer Corcoran estimated that the inaccuracy rate, based on his own experience, is as high as 20%, an inaccuracy rate that is unquestionably substantial. (See, generally, Note, Garbage In, Gospel Out: Establishing Probable Cause Through Computerized Criminal Information Transmittals, 28 Hastings LJ 509.)
Absent satisfactory explanation, and here there was none at all, a three-month failure to correct or update the police computer records is unreasonable and unacceptable. While some delay is to be expected, the People have the burden of establishing that the three-month delay in permitting the records to remain uncorrected was not the fault of the police, whether negligent or intentional. Otherwise a person, having once reported his car to be stolen and later having it recovered by the police, would be potentially subject to arrest and consequent search whenever driving in his automobile, for the indefinite future. Moreover, as demonstrated by this case, a similar fate would be lurking with regard to anyone to whom he lends the car. This situation is intolerable. To prevent it, the police must develop procedures to substantially ensure that it does not occur. This is not to say that the erroneous information would not constitute a legitimate basis to justify a stop and further inquiry. (See People v La Pene, 40 NY2d 210, supra.) But, such erroneous information, after a lapse of time reasonable to correct the computer records, is not *885sufficient standing alone, to justify an arrest and consequent search.4
To require that alarms for stolen automobiles transmitted from police computer records and similar informational sources be reliable to a substantial certainty and that if the information turns out to be completely erroneous, that the police not be at fault, does not seem to be a requirement that will hamper law enforcement authorities. On the other hand, such a requirement will, in this computer age, reduce the possibility that law abiding persons, who report that their automobiles are returned by the police authorities, will not risk arrest into the indefinite future.
Accordingly, the police conduct in this case being unlawful, the motions to suppress must be granted.
. Officer Corcoran stated that he knew of from 5 to 15 errors out of between 75 to 100 arrests based on such alarms.
. Since the substance was seized following Peterson’s arrest, which arrest is determined to have been unlawful, there is no point in detailing the circumstances of the seizure, which involved in part, cocaine that was inside a $5 bill discarded by him. (People v Howard, 50 NY2d 583.)
. The issue of the validity of an arrest based on an erroneous radio bulletin that an arrest warrant was outstanding, which warrant had actually been vacated months earlier, is before the Court of Appeals in People v Jennings (mot for lv to app granted 53 NY2d 944 [Wachtler, J.] from 81 AD2d 645, affg without opn a suppression order of County Ct, Nassau County [Thorpe, J.]).
. The developing “good faith” exception, to the exclusionary rule, e.g., United States v Williams (622 F2d 830, cert den 449 US 1127), whatever its viability in New York (cf. People v Adams, 53 NY2d 1), is not applicable to this case since the arresting officer’s conduct was unreasonable. Having had personal experience with an error rate of almost 20%, he failed to make adequate inquiry especially when told by the driver that the car had been borrowed from the owner. No request was made for the registration papers which, as the record indicates, were at all times in the driver’s possession although, the simple expedient of asking the driver for such documents may have quickly clarified the situation or at the very least put him on notice that further inquiry was called for, rather than an arrest. The failure to inquire if the driver had the automobile registration papers; the failure to try and check her contention that the car was not stolen; the failure to inquire as to the whereabouts of the owner; all coupled with the officer’s awareness that a significant number of such radio alarms, are erroneous, compels the conclusion that his conduct was unreasonable.