Your question isn't clear, but there are a couple of ways you could do this. I wouldn't bother messing around with the breaks, I would transform the data or augment it so that the breaks look after themselves. That is, I would keep the breaks and the data consistent.
library(ggplot2)
library(scales)
df <- data.frame(FY = c("FY09 Q4", "FY09 Q3", "FY09 Q2", "FY09 Q1"),
views = c(15881100, 8888000, 24168000, 17611000))
df$views.k <- df$views/1000
# Transform y in the ggplot call...
g <- ggplot(df, aes(x = FY, y = views/1000)) +
geom_bar(stat = "identity") +
scale_y_continuous(labels = comma)
print(g)
# ...or just use views.k
g <- ggplot(df, aes(x = FY, y = views.k)) +
geom_bar(stat="identity") +
scale_y_continuous(labels = comma)
print(g)