I'm not a lawyer, but I am professionally interested in this weird branch of the law, and it seems like EFF's staff attorney went a bit out on a limb here:
* Fizz appears to be a client/server application (presumably a web app?)
* The testing the researchers did was of software running on Fizz's servers
* After identifying a vulnerability, the researchers created administrator accounts using the database activity they obtained
* The researchers were not given permission to do this testing
If that fact pattern holds, then unless there's a California law governing this that I'm not aware of --- and even then, federal supremacy moots it, right? --- I think they did straightforwardly violate the CFAA, contra the claim in their response.
At least three things mitigate their legal risk:
1. It's very clear from their disclosure and behavior after disclosing that they were in good faith conducting security research, making them an unattractive target for prosecution.
2. It's not clear that they did any meaningful damage (this is subtle: you can easily rack up 5-6 figure damage numbers from unauthorized security research, but Fizz was so small and new that I'm assuming nobody even contemplating retaining a forensics firm or truing things up with their insurers, who probably did not exist), meaning there wouldn't have been much to prosecute.
3. Fizz's lawyers fucked up and threatened a criminal prosecution in order to obtain a valuable concession fro the researchers, which, as EFF points out, violates a state bar rule.
I think the good guys prevailed here, but I'm wary of taking too many lessons from this; if this hadn't been "Fizz", but rather the social media features of Dunder Mifflin Infinity, the outcome might have been gnarlier.