The story everyone is talking about today is the HIV "cure" - the Mississippi baby who, after being blasted with a cocktail of anti-viral drugs at birth, is now, at two years old, apparently virus-free. But when reading the euphoric news stories about it here , here  and here , you should bear the following in mind:
1. The baby may not even have been infected in the first place
Here's a weird section in the NYT version of the story . Have a read of the following two paragraphs - are the doctors certain or uncertain that the baby was infected?
“The one uncertainty is really definitive evidence that the child was indeed infected,” said Dr. Daniel R. Kuritzkes, chief of infectious diseases  at Brigham and Women’s Hospital in Boston.
Dr. Persaud and some other outside scientists said they were certain the baby — whose name and gender were not disclosed — had been infected. There were five positive tests in the baby’s first month of life — four for viral RNA and one for DNA. And once the treatment started, the virus levels in the baby’s blood declined in the pattern characteristic of infected patients.
The tests are pretty good, but are not usually trusted as a basis for confirmed diagnosis at that early stage. It is normal practice to confirm positive tests at 6 weeks. But as this baby had already been treated by then, lowering its viral load (negative tests came back at 29 days), it would have been difficult to do this. There is a very small chance the baby was not infected.
The virus may not have yet taken a hold on the baby's cells in a permanent way. Here's the WSJ :
Cells in the baby "may have been infected—there was virus around," said Steven Deeks, an AIDS researcher at University of California at San Francisco. "But the cells being infected weren't the type that become long-lived reservoirs."
There is also a small chance the baby was immune to HIV anyway. Around 1 per cent of Caucasians in the US are naturally immune.
Now, these are small chances, but then this baby is an outlier. It was not part of a large study where such anomalies are ruled out. The scientists have said it is unlikely to be replicable. Any way you put it, the baby itself is an anomaly.
2. This "cure" has already been found, and has been used on newly-infected people since 1987. Even if the baby was infected, today's news would simply extend it to newborns.
Today's news applies only to babies. Newborns at that. And it's not really a new method - rather, it's the same idea as PEP : if you have only just been infected, you might avoid HIV if you are immediately given a large amount of anti-HIV drugs. What the scientists confirmed today is that, in terms of emergency treatment, newborns are the same as newly-infected people.
Here's the NYT again: 
“That goes along with the concept that, if you treat before the virus has had an opportunity to establish a large reservoir and before it can destroy the immune system, there’s a chance you can withdraw therapy and have no virus,” said Dr. Anthony S. Fauci, the director of the National Institute for Allergy and Infectious Diseases.
(Oh, and if you were wondering, someone has actually been cured of AIDs before. This was a man called Timothy Brown , a leukemia sufferer who received a bone-marrow transplant from a donor genetically resistant to HIV.)
3. This would not be breaking much ground in preventing HIV in newborns anyway, because we have a solution for that
In countries with access to top-notch medical care (ie western countries), the transference of HIV from mother to child is extremely rare. This is because mothers are treated with antiretroviral therapy during pregnancy - a very effective way of preventing HIV in newborns.
4. Newborns in countries without a solution for that probably wouldn't get this treatment anyway
In countries without access to top-notch medical care, there is no reason that this treatment would be available where antiretroviral pregnancy treatment isn't.
So what does the news mean? Well it means that a few babies in countries with access to this sort of care but whose mothers have somehow slipped through the net of normal practice can be saved. Joyous news. But not quite as joyous as everyone seems to be making out.
UPDATE: This blog originally said that HIV tests at birth couldn't be accurate, but this applies only to antibody tests. The tests on DNA/RNA, which were done in this case, have a greater (although not 100 per cent) degree of accuracy.