Academic publications are sometimes subject to an embargo, in which information about the publication is shared with the media, but they are restricted from publishing a news article using this information before a certain date.
In this context, leaks are not uncommon. Sometimes the author of the paper is responsible for the leak; on other occasions, the publisher or a news agency may be responsible.
Here are two examples in which the Associated Press broke an embargo:
In both of these cases, there were no serious consequences for anybody involved.
In a case that did involve consequences, a researcher published a paper using data that was made available through the NIH, but was subject to a data embargo. The paper was retracted (see the story on RetractionWatch), and the researcher's access to the shared data was suspended:
Upon learning of the violation, the investigator’s access to dbGaP [database of Genotypes and Phenotypes] was immediately suspended pending an investigation by the NIH Data Access Committee with responsibility for the dataset involved and a review by the GWAS Senior Oversight Committee (SOC). Information pertaining to the incident was requested from the investigator’s home institution through the Institutional Signing Official that approved the investigator’s original request to the NIH. After a thorough review of the circumstances pertaining to the violation, the SOC revoked access to all dbGaP data for a period of six months.
All work with data downloaded before the date of the access suspension was expected to cease during the ban. This ban included the Primary Investigator as well as those individuals working with the individual-level GWAS data under his Data Access Request, because they also agreed to abide by the terms and conditions for data use within the Data Use Certification agreement. The period of the ban passed on March 4, 2010, and Dr. Zhang may now submit new requests for access to dbGaP data.
Another data leak scenario was mentioned by Stephan Kolassa in a comment: when a researcher uses private data that is leaked. Here is an example of a case where
the world's largest futures exchange has accused the top U.S. derivatives regulator of illegally sharing sensitive market data with outside researchers who then used the information to publish academic papers about high-frequency trading
Possibly the most high-profile case of data leakage in the very recent past is Climategate, but that turned out to be more of a "hack" than a leak by someone who had legitimate access to the data. In 2009, a server at the University of East Anglia (UEA) was compromised, and material was leaked including more than 1,000 emails, 2,000 documents, and source code, pertaining to climate change research. Many of the emails concerned technical aspects of climate research, such as data analysis. In this case, climate change skeptics argued that the leaked materials showed that scientists engaged in a conspiracy to manipulate climate data. However, various investigative committees reported no evidence of fraud or scientific misconduct on the part of the scientists.
The individuals responsible for the breach were not identified. Police said said, however, that despite rumors to the contrary, the attack had been carried out "remotely via the internet" and that there was "no evidence to suggest that anyone working at or associated with the University of East Anglia was involved in the crime."
