By CHARLIE SAVAGE

Published: August 23, 2011

WASHINGTON — Agents of the Federal Bureau of Investigation have been more likely to be hunting for potential threats to national security than for ordinary criminals in recent years, but much of the time found neither, according to newly disclosed internal information.

Data from a recent two-year period showed that the bureau opened 82,325 assessments of people and groups in search for signs of wrongdoing. Agents closed out most of the assessments, the lowest-level of F.B.I. investigation, without finding information that justified a more intensive inquiry.

Separately, the bureau also initiated 1,819 assessments during the period to identify any possible threats within particular geographic districts. That activity ranged from looking for the presence of particular organizations, like gangs or terrorist groups with definable characteristics, to evaluating other potential vulnerabilities, like a university with classified research and many foreign students.

The data, obtained by The New York Times under the Freedom of Information Act, offers a panoramic view of the bureau’s activities toward the end of a decade-long effort to transform the F.B.I. from a law-enforcement agency focused on solving crimes to a domestic intelligence agency whose mission is to detect potential threats before they can reach fruition.

The disclosure, covering March 25, 2009, to March 31, 2011, focused on assessments, which an agent may open “proactively or in response to investigative leads” and without first having a particular factual basis for suspecting a target of wrongdoing, according to the F.B.I. manual. Former Attorney General Michael Mukasey issued guidelines for the bureau creating that category in 2008.

During an assessment, agents may use a limited set of techniques, including searching databases about targets, conducting surveillance of their movements and sending a confidential informant to an organization’s meetings. But to use more intrusive techniques, like secretly reading e-mail, agents must open a more traditional “preliminary” or “full” investigation. Such inquiries require agents to first have a greater reason to start scrutinizing someone: either an “information or allegation” or an “articulable factual basis” indicating possible wrongdoing.

According to the data, during the 2009-11 period agents opened 42,888 assessments of people or groups to see whether they were terrorists or spies. A database search in May 2011 showed that 41,056 of the assessments had been closed. Information gathered by agents during those assessments had led to 1,986 preliminary or full investigations.

The data also showed that agents initiated 39,437 assessments of people or groups to see whether they were engaged in ordinary crime. Of those, 36,044 had been closed, while 1,329 preliminary or full investigations had been opened based on the information gathered.

Michael German, a former F.B.I. agent who is now with the American Civil Liberties Union, said the high number of assessments relative to the number that developed into more intensive investigations was cause for concern. He noted that the F.B.I. retained the data it collects about a target, even if the person or group turns out to be innocent.

“It’s clear the F.B.I. is casting its investigative net too broadly,” Mr. German said. “And remember that only a small proportion of ‘preliminary’ investigations become ‘full’ investigations, and only a small percentage of full investigations result in criminal charges. This data makes it clear that the ‘assessment’ authority granted in the attorney general guidelines is far too broad.”

But Valerie E. Caproni, the F.B.I. general counsel, said that the data showed that agents had been able to dispose of about 96 percent of the low-grade reasons they might have had for suspecting someone of wrongdoing, like a vague tip or some other ambiguous lead, using “low intrusion techniques” rather than by opening a potentially more invasive preliminary investigation.

The new investigation standards, Ms. Caproni said, “end up being privacy protective because previously, without a well-developed, robust assessment category, many if not most of those would have been opened as preliminary investigations.”

The newly disclosed data roughly matched a far more limited disclosure earlier this year about assessments of people and groups generated in late 2008 and early 2009, but the latest data covered a longer period and included additional detail. The data also bolsters the F.B.I.’s assertion that since the Sept. 11 attacks, the bureau has taken low-grade tips about national security threats more seriously than similarly vague and seemingly implausible leads about possible criminal activity. The 39,437 criminal assessments were based on 73,303 complaints received by the F.B.I., indicating that about half of the time, agents merely filled out a complaint form but saw nothing worth following up on.

In contrast, the disclosure did not offer a separate number of national security complaints, suggesting that some were not followed up with an assessment. The bureau says its policy calls for every national security-related tip, no matter how dubious, to be investigated. Still, Ms. Caproni cautioned that the assessment data did not offer a complete view of the F.B.I.’s activities. For example, she said, if the Central Intelligence Agency told the bureau that an overseas source had provided a specific claim about a terrorist cell operating inside the United States, agents would immediately open a more intensive investigation rather than starting with an assessment.

She also cautioned that some details about the numbers might be fuzzy. In the past, for example, the F.B.I. has noted that one assessment may uncover information that leads to more than one preliminary investigation. At the same time, agents may occasionally have neglected, when opening a preliminary investigation, to record the fact that it developed out of an assessment.

The new data also showed that while the agency has opened fewer assessments looking for broader intelligence about possible threats and vulnerabilities within a district, a category it calls Type 3 assessments, it has been slower to close them. Of the 1,819 it had opened during the period, 1,056 were open by May.

Those figures covered Type 3 assessments generated both by national security programs — like domestic and international terrorism, weapons of mass destruction and counterintelligence — and criminal programs, like civil rights, gangs, organized crime, violent crime and white-collar crime. The F.B.I. censored the specific breakdown for each category. The data release comes as the F.B.I. is preparing to issue agents a new version of its manual, the Domestic Investigations and Operations Guide, that will relax some rules about the techniques allowed at the assessment stage. Ms. Caproni said the new rulebook is at the printer now, and a partly redacted version is likely to be released to the public at the same time it takes effect around mid-October.

http://www.nytimes.com/2011/08/24/us/24fbi.html?_r=1&hpw