Police officers may be more likely to suffer assaults if they are wearing body-worn cameras, a new study suggests.
Preliminary results from eight UK and US forces indicated that rates of assault were 15% higher when personnel used the devices.
Experts said one possible reason for the "unexpected" finding could be that officers feel more able to report incidents once they are captured on camera.
The research also suggested there was no overall discernible effect of using the cameras on police use of force on citizens.
Most police services in England and Wales now use body-worn video to some extent, and last year the largest force - the Metropolitan Police - announced plans to introduce the equipment to all frontline officers following a large-scale trial.
For the new study criminologists worked with eight police forces across the US and UK - including West Midlands, Cambridgeshire and Northern Ireland's PSNI - to conduct ten randomised-controlled trials.
Over the 10 trials, the experts from the University of Cambridge and research organisation RAND Europe, found that rates of assault against officers wearing cameras on their shift were an average of 15% higher, compared to shifts without cameras.
The authors wrote that the result "demands attention".
They said: "Does this mean that officers should be advised to remove BWVs (body-worn videos) immediately?
"If the results are accepted uncritically, that is, that BWVs increase the likelihood of assaults against officers, then this might be the conclusion one comes to. However, we cannot rule out alternative explanations at this stage."
A number of theories were advanced, including the idea that "with an 'objective' record of events, officers feel more able (or compelled) to report instances when they are assaulted".
Another possibility was that officers may be "less assertive" because of monitoring, and this could make them more vulnerable to assault, the study suggested.
It also noted that the strongest results for assaults against police came from the smallest studies and these may be "atypical results", while increased assaults against police "may also be a corollary of the decreased use of force in some instances".
Shift patterns of more than 2,000 officers were split at random between those allocated a camera and those without the kit. A total of 2.2 million officer-hours were covered.
In another finding that was said to run contrary to current thinking, the rate of use-of-force by police on citizens was unchanged by the presence of body-worn cameras.
Deeper analysis showed this depended on whether or not officers chose when to turn cameras on. If they were turned on and off during their shift then use of force increased, while if they were kept rolling for the whole shift it decreased.
Dr Barak Ariel, of the University of Cambridge's Institute of Criminology, said: "If an officer decides to announce mid-interaction they are beginning to film, for example, that could provoke a reaction that results in use of force.
"Our data suggests this could be what is driving the results."
The researchers urged caution on the findings-published across two papers in the European Journal of Criminology and the Journal of Experimental Criminology- saying work is ongoing and the results demand further scrutiny.
Nerys Thomas, head of research at the College of Policing, said: "The researchers for this trial are rightly urging caution in drawing conclusions from these findings.
"It seems that the results on assaults against officers are based on a very small number of incidents, as reported by officers."
The body has conducted its own research on body-worn video.
Ms Thomas added: "It is difficult to compare these to College of Policing results for body worn video because these trials are based in sites in the US and UK.
"The College of Policing has published findings from its randomised control trial with the Metropolitan Police and found where officers were wearing cameras, there was a 33% reduction in allegations against them and there was no effect on officer safety and use of force.
"In addition, it found the cameras did not alter the quality of policing and offered officers greater confidence if challenged, as well as footage to support their decision-making, for example, during stop and search."