The recently-revealed continuing leaks of radioactive contaminants from Fukushima, although they are 500 times smaller than the initial release and 100 million times smaller than the natural radioactivity of the Pacific Ocean, could still be dangerous to local ecology and human health, but does not represent a global catastrophe.
This is my conclusion as a non-expert in the field summarizing the publicly available information. It could be wrong.
Details follow.
Different accounts give different amounts of radioactive water leaked into the Pacific from the Fukushima nuclear plant via continuing leaks. Some of them give the amount, rather uselessly, in tons, but the better accounts give the amount of radioactive material in the water in becquerels. One TV report says it's a PBq, but that sounds like it's probably referring to part of the initial, much larger release; other reports give the amount as 10 to 50 TBq: More Fukushima Fallout and a Japan Times article say 30 TBq, Asahi Shimbun's article says 24 TBq, while National Geographic's article gives the groundwater concentration of radioactive cesium in places as around 1kBq/kg, gives the total release as 0.3 TBq/month, describes the immediate aftermath of the disaster as a release of around 10PBq, and contextualizes it by comparing to the 89 TBq release of cesium-137 from the Hiroshima bombing.
The raw becquerel numbers are sort of meaningless without context on how big a becquerel is. Mijlkovic's 2012 article says:
9.3 percent of the catches exceeded Japan’s official ceiling for cesium, which is 100 becquerels per kilogram (Bq/kg). ... Canada’s much higher ceiling, which is 1,000 Bq/kg
Comparing to natural radioactivity, typical rocks and dirt have hundreds of Bq/kg, mostly due to potassium, but with significant amounts due to uranium, thorium, and radium. Seawater has on the order of 10 Bq/kg, almost all due to potassium. A human body contains on the order of 10 kBq, mostly due to potassium and carbon-14. The Pacific Ocean naturally contains about 7000 EBq, because it contains about 700 million km³ of water.
That means that, purely in terms of increased radioactivity, the leak is so small as to be insignificant. A 70 TBq leak of radioactivity would be 100 000 000 times smaller than the natural radioactivity of the Pacific Ocean. Increasing the radioactivity by one part in a hundred million is not dangerous; if all radioactivity were equivalent, it wouldn't even be detectable. Even the initial 10PBq release was only enough to increase total Pacific Ocean radiation by about one part per million.
The fatal dose of cesium-137 is on the order of 100MBq/kg for dogs. So a 70TBq leak, if not sufficiently diluted, is enough to kill about 700 000 kilograms of dogs or other similar animals, such as people — about ten thousand people.
However, it's sensible to have different safety limits for different radioactive elements, because some of them, like potassium, are biologically regulated at a constant level — so eating more radioactive potassium probably doesn't increase your exposure to radiation at all — while others bioaccumulate, like strontium-90, which replaces calcium in your bones and can therefore continue irradiating you for the rest of your life. Cesium bioaccumulates to some extent, so it becomes more concentrated in animals than in plants, more concentrated in predators than in herbivores, and more concentrated still in secondary predators like tuna. Tritium, a third radioactive contaminant in this case, is not known to bioaccumulate, but it also isn't homeostatically bioregulated like potassium; chemically, it's almost identical to hydrogen.
Also, the radioactive material is not evenly distributed. Ocean-caught fish from near the Fukushima reactor had levels up to tens of kBq/kg in 2012, hundreds of times higher than normal fish, and we can reasonably expect that fish that feed in areas where the water has been released will continue to be contaminated to much higher than normal levels, perhaps 100 times higher than normal. However, it's believed that the water that has leaked has already had most of its radioactive cesium removed, unlike the water that leaked early on; strontium-90 may be a bigger concern now.
So there are real health concerns, but they are not very large with the current size of the leak.
The National Geographic article says that the total amount of contaminated water stored, now and in the future, is on the order of a million tons; only a third of that is there now. Unfortunately, I don't have a good handle on how many becquerels that water contains. It appears that the total water loss was 300 tons, mixing into 400 tons per day of groundwater; that gives us roughly 30 TBq/300 tons or 11 MBq/kg, about a million times more radioactive than natural seawater. A million tons of 11MBq/kg water would add up to 10PBq, roughly the same size as the initial Fukushima disaster. So, in a worst-case scenario where all the tanks vented into the ocean, it would be roughly comparable to the initial incident, and despite bioaccumulation of strontium, probably would not be enough to cause more than local ecological damage.