A quick scan of topics seems to indicate that I'm the one to break this story here. Some government agencies in the UK are using voice stress analysis (VSA) on callers claiming government benefits, such as welfare and housing assistance, in an attempt to reduce fraud. Cases which are flagged as indicating possible deception are followed up with more in-depth investigation.
Here is a link to a quick CNN video on the story which I recommend.
I find this interesting for a number of reasons. First, the American Polygraph Association is a very harsh critic of voice stress technology, most particularly Computerized Voice Stress Analysis, a proprietary brand of voice stress analysis peddled by the "National Institute of Truth Verification" (NITV) a Florida-based company. Indeed, they have a
large collection of information about how inaccurate it is on their website and decry the fact that over 1400 American police departments use the technology. There is, of course, a great deal of self interest in their position, since CVSA is often used in lieu of the polygraph: they figure the fewer departments that use CVSA the more that will use the polygraph, giving them more business. Nonetheless, their collection of anti-cvsa information is possibly the best on the web. Too bad they don't make available similarly good information that is pro-polygraph.
Second, the technology might be working. No, I don't think it is reliably detecting deception, but that's not what it is supposed to be doing: it is supposed to be cutting down on fraud. Before being connected with an operator, callers hear a recording that indicates that VSA is being used (and, I presume, touts it as highly accurate and indicates the stiff penalties of fraud). In the CNN story linked to above, a government spokesman says that prior to using the technology, only 10% of callers reported a change in status that would decrease benefits (a new job, a new roommate to help pay bills, etc) but after implementation of the program, that number jumped to 35%. Assuming that the 25% jump isn't accounted for by an upturn in the economy or normal seasonal variation, it might well be true that the VSA is cutting down on benefits fraud and saving taxpayers significant money.
It would be interesting to have an impartial group examine the data to see if it actually is the case that the program is cutting down on government fraud and, if so, if the savings exceed the costs of implementing the program. I suspect it will be hard to accomplish that, as the bureaucrats and VSA pushers will strongly resist doing so, since they risk losing money and power if the program is not found useful.
There are quite a few differences between this use of the VSA and the common use of the polygraph in this country, however. There appear to be no real negative consequences for false positives (of which there are no doubt many) for benefits claimants beyond some inconvenience in providing additional records and information to investigators. Also, it is not clear how much officials actually believe their own lies about the accuracy of the device. One official said it detects deception "in excess of 95 per cent, if not higher," which I can't possibly imagine he actually believes. So my impression is that they realize it's a crock, but a useful one for saving money. If there is no pre-existing polygraph industrial complex in the UK and an existing polygraph culture, it is less likely that they believe the claims. If they keep in touch with reality, the dangers of the program are minimized.
I found a couple of other stories about this (try a google news search) but none provided more than a brief overview. It'll be interesting to see if more information about this program comes out.