Efficient Market Hypothesis as Law
There seems to be two intellectual objections to the 'fraud on the market' principle adopted by the the US federal courts and the securities regulators (e.g., the Securities and Exchange Commission). The first type of objection is that, assuming the efficient market hypothesis should be used as the guiding framework to analyze the workings of the financial markets, the courts are misapplying it by not distinguishing between the issue of relying on the distorted pricing information and the issue of what type of damages or other punishments should be associated with the distorting behavior. This argument is laid out in Bradford Cornell's paper, Market Efficiency, Crashes and Securities Litigation.
The second type of objection is that the markets aren't really that efficient any way; i.e., the efficient market hypothesis (even in a relatively weak form) shouldn't be used as the analytical framework by the courts. I should note that, even if this second type of objection should be the better economics (although not if you're from a Chicago School mindset), it is not necessarily good law since it seems to me that insiders that misuse insider information to illegallly profit should be found guilty and/or liable for committing a fraud on both the markets as well as the existing shareholders.
To paraphrase James Buchanan, a Nobel prize winning economist, law and economics don't always mix.