dissenting.
Because there is no evidence in the record sufficient to infer that the owner of the dog, Supan, knew or should have known of the dog’s propensity to bite a human being, the trial court erred by denying Supan’s motion for summary judgment.
For the Griffins to recover damages from Supan resulting from the dog bite suffered by Bo Griffin, they were required to prove under the “first bite” rule that Supan knew or should have known of his dog’s propensity to do the particular act which caused the complained of injury. Hamilton v. Walker, 235 Ga. App. 635 (510 SE2d 120) (1998); Durham v. Mooney, 234 Ga. App. 772, 773 (507 SE2d 877) (1998). The particular act which caused injury in this case was biting a human being. This does not mean it was necessary to prove that Supan knew or should have known of his dog’s propensity to do the exact same act that caused the injury at issue, “but in order to infer the requisite knowledge there must be at least one incident that would cause a prudent person to anticipate the actual incident that caused the injury.” Id. at 773; Torrance v. Brennan, 209 Ga. App. 65, 67-68 (432 SE2d 658) (1993). Although the prior incident known to the owner need not have been exactly like the incident at issue, it nevertheless must be of the same type as the incident at issue before the requisite knowledge of the incident which caused the injury will be ascribed to the owner. Id. at 68. “[T]his Court consistently has held that the dog must have, on a prior occasion, done the same act which resulted in the injury comprising the tort action.” Hamilton, 235 Ga. App. at 635. For example, where a dog was known to chase people, but had never chased a car or motorcycle, there was a lack of evidence to support an inference that the owner knew or should have known that the dog had a propensity to chase a motorcycle and cause it to crash. Id. at 635. Similarly, where a dog was known to display menacing or aggressive behavior toward human beings, but had never bitten a human being, knowledge of the dog’s propensity to bite a human being will not be ascribed to the owner. Banks v. Adair, 148 Ga. App. 254, 255 (251 SE2d 88) (1978); Hamilton, 235 Ga. App. at 636; Durham, 234 Ga. App. at 773.
In support of his motion for summary judgment, Supan filed an affidavit stating that he had no knowledge of the dog’s propensity to bite a human being. In opposition to summary judgment, the Griffins filed an affidavit from Steven Keller, a neighbor of Supan, who stated that three months prior to the day Bo Griffin was bitten, four or five of Supan’s dogs (including the dog that bit Bo Griffin) came to his residence, attacked his dog, and threatened him with “bared fangs, vicious growls and attack behavior.” Keller further stated that, on the day this incident occurred, he told Supan of the incident and told Supan that his dogs were vicious and should be locked up. Keller said *408that Supan “acknowledged that the dogs were a problem and told [him] that if the dogs ever came back on [his] property, to do whatever was necessary in [his] opinion to keep the dogs from attacking and off [his] property.” Keller did not claim that the dogs actually attacked or bit him, and there is no evidence that this occurred.
The majority concludes that Keller’s statement that Supan told him “to do whatever was necessary in [his] opinion to keep the dogs from attacking” creates a jury issue “as to Supan’s prior knowledge of his dogs’ tendency to attack humans.” There is absolutely no evidence in this case to support an inference that Supan knew or should have known that the dog at issue had a propensity to attack or bite a human being. The dog did not attack or bite Keller during the incident described by Keller in his affidavit, nor is there any evidence that the dog had attacked or bitten anyone prior to biting Bo Griffin. Supan’s statement to Keller that the dogs were a problem and that Keller should do whatever was necessary to keep them from attacking showed nothing more than Supan’s acknowledgment of the incident which Keller had just described to him in which the dogs attacked Keller’s dog and displayed menacing behavior toward Keller by baring their fangs and growling in what Keller characterized as “attack behavior.”
At most, Supan’s statement to Keller showed that Supan knew the dogs had displayed menacing behavior which Keller characterized as a threat to attack him and showed Supan knew the dogs had actually attacked Keller’s dog. To interpret Supan’s statement as evidence that he had knowledge of the dog’s propensity to attack or bite a human being — without any evidence that the dog had previously attacked or bitten anyone — is sheer sophistry. It creates an inference based on pure speculation, conjecture or possibility, which this Court has recognized is insufficient to create a genuine issue of fact. Heinsimer v. Wellington Leisure Products, 231 Ga. App. 579, 582 (500 SE2d 7) (1998). Moreover, in opposition to the inference drawn by the majority, Supan gave positive, unrebutted testimony that he had no knowledge of the dog’s propensity to bite a human being. In considering a motion for summary judgment, a fact which may be inferred, but is not demanded, by circumstantial evidence has no probative value against positive, unrebutted evidence that no such fact as sought to be inferred exists. Id. at 582; Beeson v. Crouch, 227 Ga. App. 578, 580 (490 SE2d 118) (1997).
The only other basis for the inference drawn by the majority that Supan had knowledge of his dog’s propensity to attack a human being is the evidence that Supan knew his dog had displayed menacing behavior toward Keller, characterized by Keller as attack behavior, and had attacked Keller’s dog. But to draw the inference on this basis contradicts long-standing “first bite” precedent in which this *409Court has held that neither a dog’s known menacing behavior toward human beings nor its known attacks on other animals are sufficient to allow an inference that the owner had knowledge of the dog’s propensity to attack or bite a human being. A dog’s attacks on other animals are insufficient to put an owner on notice that his dog has a propensity to bite or attack a human being. Hamilton, 235 Ga. App. at 636. This Court has also consistently held that an owner’s knowledge that his dog has previously displayed menacing or aggressive behavior toward human beings is not sufficient to support an inference that the owner knew or should have known that the dog had a propensity to attack, bite, or injure a human being. Banks, 148 Ga. App. at 255; Hamilton, 235 Ga. App. at 636; Durham, 234 Ga. App. at 773; compare Johnson v. Kvasny, 230 Ga. App. 162, 163 (495 SE2d 651) (1998) (where the dog owner admitted after the bite at issue that she “knew that something like this would happen”). For example, in Starling v. Davis, 121 Ga. App. 428, 429 (174 SE2d 214) (1970), despite known incidents in which the dog had chased but not bitten people, and an incident in which one person had taken refuge on a car to escape the dog, we concluded this was not evidence that the dog owner had knowledge that the dog had a propensity to bite a human being. In Wells v. Beach, 169 Ga. App. 736, 737 (315 SE2d 23) (1984), evidence showed that the dog had never bitten anyone, but had displayed menacing behavior toward a child, that the child had slapped the dog to keep it from biting him, and the dog pawed at the child, causing the child to fall down and suffer a concussion. Nevertheless, we concluded this known behavior was insufficient to establish the dog owner’s knowledge of the dog’s propensity to bite a human being. In Hamilton, 235 Ga. App. 635, and Durham, 234 Ga. App. 772, although the dogs at issue had growled, barked and displayed threatening and menacing behavior toward people, they had never bitten anyone, and we found under the “first bite” rule that this known behavior was not sufficient to support an inference that the owners knew or should have known that the dogs had a propensity to bite a human being. Moreover, the majority’s repeated references to the fact that the dog was a “Rottweiler and Chow mix,” apparently in an effort to bolster the inference that Supan knew or should have known of the dog’s propensity to bite, are completely irrelevant. This Court has held that a dog’s breed and size provide no evidence that the owner had knowledge of the dog’s propensity to bite. Stanger v. Cato, 182 Ga. App. 498 (356 SE2d 97) (1987).
The majority cites McBride v. Wasik, 179 Ga. App. 244 (345 SE2d 921) (1986) as controlling authority in support of the inference that Supan knew or should have known that his dog had a propensity to bite a human being. In McBride we concluded there was evidence that the dog owner had knowledge of his dog’s propensity to bite a *410human being even though there was no evidence that the dog had previously bitten anyone. We reached this conclusion, however, on facts that bear no resemblance to the facts in the present case. Although the dog which bit the victim in McBride had never previously bitten or attacked a person, it was a “trained attack dog.” Id. There was evidence that the dog’s owner had previously commanded the dog to attack the victim’s wife, yelling, “Kill the bitch!” and that the dog charged on this command but was called off by the owner before it reached the victim’s wife. Id. Additional evidence showed that, shortly before the victim was bitten, the owner had said that “he hoped that dog got out and went over there and killed [the victim].” Id. at 245. The dog had also previously charged and leaped upon numerous small children, knocking them to the ground. Id. at 244. The evidence in McBride showed far more than an owner who, like Supan, knew only that his dog had attacked another animal and engaged in menacing or aggressive behavior toward a person. It showed an owner who knew his dog was specifically trained for and had the ability and willingness to attack and injure a human being. This was enough, without a previous bite or attack on a human being, to support an inference that the owner knew or should have known that the dog had a propensity to bite a human being. Id. at 245. See also Sanders v. Bowen, 196 Ga. App. 644, 645-646 (396 SE2d 908) (1990) (a reasonable inference could be made that an owner of a dog trained to be vicious, aggressive and attack-oriented so it could fight for money, knew or should have known of the dog’s propensity to bite a human being even though it had not previously bitten anyone); Torrance, 209 Ga. App. at 66-67 (where the dog had previously put its mouth completely around a person’s arm leaving red marks, had jumped on another person and used its teeth to rip the person’s pants, and had “nipped” another person on the buttocks, this was sufficient, without a previous bite, to support a verdict finding that the owner had knowledge of the dog’s propensity to bite a human being). Unlike these cases, Supan’s dog was neither a trained attack or fighting dog, nor had the dog previously grabbed people with its mouth, “nipped” people, or ripped people’s clothes with its teeth. Supan knew only that his dog had attacked Keller’s dog and had engaged in aggressive and menacing behavior toward Keller.
Finally, the majority delivers a coup de grace to the “first bite” rule in the context of a premises liability claim against a dog owner. The majority concludes that the victim, Bo Griffin, who was bitten at Supan’s residence, was a “Good Samaritan” invitee (as opposed to a visitor-licensee) at the residence. The majority then holds that: “With proof regarding Lavern Supan’s prior knowledge of his dog’s vicious tendencies, the true test of liability in the case sub judice is Lavern Supan’s superior knowledge of his dog’s temperament.” This test dis*411cards the long-standing “first bite” rule, which required evidence of the owner’s knowledge that, on a prior occasion, the dog had bitten a human being or done the same type of act before liability could attach for the second bite. In its place, the majority’s new test allows liability to attach for the first bite based on the owner’s prior superior knowledge of the dog’s “vicious tendencies” or “temperament.” In other words, dog owners like Supan, whose dogs have never bitten or attacked anyone, but who know their dogs have displayed menacing or aggressive behavior toward a person — behavior which might be subjectively characterized as a display of a vicious tendency or temperament — are now subject to liability for the first bite.
The majority opinion has the effect of implicitly overruling longstanding “first bite” precedent. Under the “first bite” rule, regardless of whether the Griffins’ claim is construed as made under the “dog bite” statute, OCGA § 51-2-7, or under the premises liability statute, OCGA § 51-3-1, the result is the same. Under OCGA § 51-2-7, there is no evidence to support a finding that Supan knew or should have known that his dog had a propensity to bite a human being. Durham, 234 Ga. App. at 773; Hamilton, 235 Ga. App. at 635. Under OCGA § 51-3-1 (applying the “superior knowledge” rule in conjunction with the “first bite” rule), there is no evidence to support a finding that Supan knew or should have known his dog had a propensity to bite a human; therefore, Supan had no superior knowledge making it foreseeable to him that the dog would bite a human being. Stanger, 182 Ga. App. at 499; Hackett v. Dayton Hudson Corp., 191 Ga. App. 442, 444 (382 SE2d 180) (1989).1 In the absence of evidence to support a cause of action under either statute, Supan was entitled to summary judgment. Lau’s Corp. v. Haskins, 261 Ga. 491 (405 SE2d 474) (1991).
The rule under which dog owners are protected from liability for their dog’s “first bite” sets a reasonably clear standard by which owners can gauge the risk of second bite liability posed by owning a dog known to have engaged in prior acts of biting or like conduct. The test of liability set forth in the majority opinion creates a vague standard under which first bite liability may be imposed on dog owners whose dogs had never previously engaged in biting or like conduct, but whose dogs had engaged in known conduct which might be subjectively characterized as indicative of a vicious tendency or temperament. Under the majority opinion, every dog owner in the State whose dog has growled and bared its fangs may now be subject to first bite liability based on a claim that the dog has a known vicious *412tendency or temperament. This test dramatically increases the risk of liability arising from dog ownership while making it virtually impossible for dog owners to determine when their dog’s conduct has placed them at increased risk. For these reasons, I dissent.
Decided June 2, 1999 — Cert. applied for. Tisinger, Tisinger, Vance & Greer, Glenn M. Jarrell, Kenneth B. Crawford, for appellant. Patrick J. Araguel, Jr., for appellees.The analysis would be the same under the facts of this case whether the victim is considered to be a licensee or an invitee. Webb v. Danforth, 234 Ga. App. 211, 212 (505 SE2d 860) (1998); see Hannah v. Hampton Auto Parts, 234 Ga. App. 392, 393-394 (506 SE2d 910) (1998).