If you're still trying to wrap your head around the saga between OpenAI, Anthropic, and the Department of Defense, you're not alone. OpenAI's move to strike a deal with the Pentagon, and its no-so-easy-to-follow justification for doing so, which Fortune's Sharon Goldman scooped on Friday afternoon, was part of an extraordinary sequence of events.

To judge by OpenAI CEO Sam Altman's decision to hold an AMA with the public about it on Saturday, it seems that many people are still confused—and a little uneasy—about the deal OpenAI made, how it differs from Anthropic's ill-fated deal, and about the implications of the DoD tactics.

Speaking of the DoD tactics, Fortune's Lily Mae Lazarus has a nice look at Emil Michael, the former (and controversial) Uber executive who posted a now infamous tweet calling Anthropic's CEO a liar with a 'God complex.'

And in another interesting twist, the Wall Street Journal reports that Anthropic's Claude was used by the U.S. DoD in the Iran attack that began just hours after the government declared it would stop using its AI technology.

This story is developing, as they say, but Fortune AI Editor Jeremy Kahn has a good overview of what makes it so important and unprecedented.—AO

As U.S. and Israeli attacks degrade Iran’s conventional military capabilities, cyber attacks will become an increasingly attractive option for Iran to strike back, security experts say.

“The Islamic Republic has always had great pride in cyber capabilities within the security services,” said Brian Carbaugh, co-founder and CEO of AI-based security firm Andesite and former director of the CIA’s elite Special Activities Center (SAC). That pride isn’t likely to evaporate with the loss of senior leadership, and may intensify as other options narrow. 

The next 48 hours are likely to be a period of “extreme volatility” where hacktivists and proxies “take the lead in escalation to fill the vacuum left by Tehran’s central command,” cyber intel firm Flashpoint said on Sunday. These actors are allegedly using Telegram and Reddit as a coordination hub, posting screenshots of alleged attacks as proof, although it takes weeks and sometimes months to verify accuracy, said Kathryn Raines, a former NSA expert who is now a threat intel team lead at Flashpoint. 

“For business leaders and those protecting businesses and making decisions at a very high level, they need to be prepared for this to continue on for some time and for the conflict to take a number of different courses of direction and swerve around the road," Carbaugh told Fortune.—Amanda Gerut

Google announced it’s developing a new data center complex south of Minneapolis featuring the world’s largest battery storage system, a big step for the nascent long-duration battery industry that could help make renewable energy-powered data centers more viable.

While most storage systems provide power in four-hour durations, or increasingly, eight hours, the iron-air battery technology Google is using, from a company called Form Energy, aims to dispatch up to 100 hours of power at a time. That means the batteries can keep the power on during prolonged severe weather events, peak summer demand, or just a particularly cloudy week that weakens solar power. 

The Minnesota data center project includes building 1.4 gigawatts of wind power, 200 megawatts of solar, and the 300-megawatt Form battery system. The 300 megawatts, scheduled for installation in 2028, are enough to power more than 200,000 homes when dispatched.—Jordan Blum