We’re heading into summer, but the chilly relationship between government bodies and private tech businesses is growing frostier by the day. In the latest development, it has emerged that Twitter requested one of its key B2B partners, Dataminr — a service that offers advanced social media analytics and early detection of major events like terrorist attacks or natural disasters — stop providing U.S. intelligence agencies with their tools and content.

But Dataminr isn’t ending its relationship with the government altogether: Dataminir still counts In-Q-Tel, the non-profit investment arm of the CIA, as an investor. Dataminr has taken investment from Twitter, too, highlighting some of the conflicts that remain as tech companies fight for more transparency and autonomy from government control.

Interestingly, the agencies who are at the center of today’s news were using Dataminr in an unpaid pilot, TechCrunch understands. That pilot, which was coming to an end, could not be continued as a paid deal because of pre-existing Twitter policies that forbid selling data for use in government surveillance.

The news of Dataminr cutting off intelligence groups was first reported by the WSJ, and we have confirmed the details directly with sources.

What exactly is Dataminr? The company uses Twitter firehose data — the unfiltered, full stream of Tweets from Twitter’s 300m+ users — along with other primary sources, to surface signals for big events as they are happening.

The company has a number of customers in the finance industry, and it works with government organizations like the Department of Homeland Security, which has a $255,000 contract with the startup.

“We have never authorized Dataminr or any third party to sell data to a government or intelligence agency for surveillance purposes,” a Twitter spokesperson told TechCrunch. “This is a longstanding Twitter policy, not a new development.”

Dataminr is not without a wider financial relationship with the intelligence agency: In-Q-Tel, the CIA’s investment arm, has a strategic stake in Dataminr, as highlighted in April in an article published in the Intercept.

Twitter is also an investor in Dataminr — with a 5% stake in the big data startup. Dataminr has raised nearly $180 million to date, with its most recent round at a $700 million valuation. Other investors include Credit Suisse, Fidelity, IVP, Venrock, among others.

Government agencies have been gathering intelligence about individuals and large-scale events from social networks more frequently as part of their efforts to keep surveillance up to speed with technology. But in the wake of Edward Snowden’s revelations about NSA surveillance, tech companies are not always comfortable playing a role in government intelligence-gathering, and are becoming more proactive about defining their boundaries.

Twitter is in the midst of a lawsuit against the U.S. Government in which it is fighting to disclose information about the requests it receives for user data from the government.  (The suit has been partly dismissed but is still in progress.)

Twitter has been celebrated in part for its role as a platform for free speech — a key virtual “town square” where people can speak their mind and organize with others in a spirit of activism. But there is also a pragmatic reason for companies like Twitter to push back against intelligence authorities using tools to monitor Twitter activity. Twitter relies heavily on consumer trust to operate: if users think that it’s a hotbed for CIA snooping, this could turn them off from tweeting.

Ironically, Dataminr may be one of the more innocuous of the social media big-data analytics firms: it uses public information from Twitter to track big events, rather than collecting and organizing personal information about people. Twitter itself pointed this out.

“Dataminr uses public Tweets to sell breaking news alerts to media organizations such as Dow Jones and government agencies such as the World Health Organization, for non-surveillance purposes,” Twitter said in a statement to TechCrunch.

Still, as the WSJ noted, it’s the optics of having any kind of relationship with intelligence agencies that motivated Twitter to end the pilot program. “American technology companies can continue to thrive only if people around the world trust them,” ACLU deputy legal director Jameel Jaffer told TechCrunch. “It’s completely understandable that a social media company like Twitter doesn’t want to be seen as an arm of American intelligence agencies.”

And in the end, it seems like cutting off Dataminr activity is not necessarily the final word. Twitter told the Wall Street Journal that the data in question is “largely public and the U.S. government may review public accounts on its own, like any user could.”

Jaffer agreed that the government could still easily observe action on Twitter. “It’s doubtful that this move by Twitter will have any significant effect on the U.S. government’s ability to access or analyze information,” he added.